Cantilever Snap-Fit Performance

Analysis for Haptic Evaluation

Jingjing Ji
Department of Mechanical Engineering, This paper investigates the parametric effects, which include material properties, hook
Zhejiang University, shape, and shear deformation, on the force=deflection relationship governing the assem-
Hangzhou 310027, China bly=disassembly processes of a snap-fit for developing embedded algebraic solutions to
achieve realistic force feedback through a haptic device. For this purpose, an algebraic
Kok-Meng Lee model, which isolates individual parametric factors that contribute to the cantilever hook
Woodruff School of Mechanical Engineering, deflection, has been derived for examining assumptions commonly made to simplify mod-
Georgia Institute of Technology, els for design optimization and real-time control. The algebraic model has been verified
Atlanta, GA 30332-0405 by comparing computed results against those simulated using ANSYS FEA workbench and
e-mail: kokmeng.lee@me.gatech.edu published approximate solutions. Additionally, the model has been validated by compar-
ing the friction coefficients of three different snap-fit designs (with same materials), which
Shuyou Zhang1 closely agree within 5% of their root-mean-square value. Implemented on a commercial
Department of Mechanical Engineering, PHANTOM haptic device, we demonstrate the effectiveness of the model as embedded alge-
Zhejiang University, braic solutions for haptic rendering in design. Nine individuals participated in evaluating
Hangzhou 310027, China a set of design options with different parameter settings; 78% of whom chose the optimal
e-mail: zsy@zju.edu.cn. theoretical solution by feeling the feedback force. These findings demonstrate that the
design confidence of assembly robustness can be enhanced through a relatively accurate
virtual force feedback. [DOI: 10.1115/1.4005085]

Keywords: snap-fit, cantilever hook, assembly, disassembly, virtual force feedback,

haptics, psychophysics

1 Introduction ver snap-fits under manual assembly for applications in the auto-
motive industry, where force=tactile feedback is essential for
Snap-fits are widely used in various types of products, equip-
sensing the full engagement of snap-fit parts during the assembly
ments, and power systems to provide attachment functionality in
of critical components (such as electrical and fuel system inter-
assemblies as they can be incorporated as feature molded in parts
connects). The findings in Ref. [10] have motivated us to explore
thus reducing part count; and their ease of assembly and disassem-
extending the method by incorporating haptic rendering, which
bly lowers manufacturing cost and time. Snap-fits performance
offers designers useful feedback force as they assemble snap-fits,
analysis requires not only the virtual simulation results but also a
to enhance product design involving snap-fits in virtual environ-
good understanding of the inherent motion=force transmission
ment. Haptic technology, which has been widely introduced to
through deformation encountered during assembly, in which a
machine, robot automation, biotechnology, and medical product
sense of human touch to feel the level of contact force is needed.
design assessment [1115], could provide feedback force to ena-
Motivated by the emerging haptic technology that helps humans
ble designers to experience the deformation in virtual environ-
feel the sense of touch, this paper develops an analytical model
ment, thereby allowing them to make objective evaluation of
for enhancing performance analysis in virtual manufacturing.
product performance. However, accurate solutions that can be em-
Traditionally, snap-fits were designed based on classical me-
bedded in haptic technology to provide real-time force=displace-
chanical knowledge; for example, Bonenberger [1] offered a com-
ment feedback in virtual design environment are required to
prehensive description for analyzing snap-fits assembly and
provide confidence of assembly robustness that can be enhanced
disassembly, in Refs. [24] linear beam theory was introduced,
in industrial settings.
and in Ref. [5] a method for attachment design concept in integral
For the above reasons, this paper examines the parametric
snap-fit assembly was proposed. Recently, finite element methods
effects (which include material properties, hook shape, and shear
(FEMs) have been widely used to obtain accurate numerical solu-
deformation) on the force=deflection relationship governing the
tions [69]. In general, FEMs are computationally time demand-
design of a snap-fit. Along with a relatively complete model
ing, which motivate engineers to concentrate on model
(CM), this investigation provides a basis for developing embedded
optimization and dimension reduction to overcome the high com-
algebraic solutions that can be efficiently implemented to achieve
putational expense. In Ref. [8], a computational model using non-
realistic force feedback through a haptic device when optimizing
linear constrained minimization was presented to efficiently
design geometry of snap-fits in real time. The remainder of this
design flexible fingers subjected to large deflection without loss of
paper offers the following:
accuracy, where the theoretical accuracy was verified by compar-
ing to simulated results against FEM. A nonlinear algebraic reduc- (1) In the context of snap-fits, a relatively complete algebraic
tion method was proposed in Ref. [9] to offer the generality of 3D model for analyzing both assembly and disassembly of a
simulation and the computational efficiency of 1D simulation. snap-fit is presented. With real-time haptic evaluation of
Snap-fits offer resistance to engagement during the assembly designs in mind, this model isolates individual parametric
and disassembly processes. Experiment fixtures were used in Ref. factors that contribute to the cantilever hook deflection.
[10] to provide the force and tactile feedback in preloaded cantile- Thus, the algebraic model provides an effective means to
examine assumptions often made to reduce models to trac-
table form for design optimization and real-time control.
1
Corresponding author. The relation between feedback force and deflection has
Contributed by the Design Theory and Methodology Committee of ASME for
publication in the JOURNAL OF MECHANICAL DESIGN. Manuscript received February 13,
been deduced to illustrate both assembly and disassembly
2011; final manuscript received August 28, 2011; published online December 9, of a snap-fit; several commonly used simplified models are
2011. Assoc. Editor: John K. Gershenson. compared.

Journal of Mechanical Design Copyright V

C 2011 by ASME DECEMBER 2011, Vol. 133 / 121004-1

Downloaded 18 Dec 2011 to 222.205.33.68. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
 Fig. 1 Cantilever hook and matching part; (a) coordinate systems and characteristic dimen-
sions, (b) sketch illustrating deflection, (c) forces components in insertion, and (d) forces com-
ponents in retention

(2) The proposed model has been verified with two methods. friction) as the cantilever deflects and reaches its maximum
This first method compares computed results against those deflection at x m .
simulated using ANSYS FEA workbench and published ap- The matching part slides on the tip while maintaining its max-
proximate solutions. The second method uses a coupled imum deflection during dwelling.
calculation=measurement approach to determine and com- The elastic deformation gradually returns to zero as the
pare the friction coefficients of three different snap-fit matching part moves down the inclined surface completing
designs (with same materials), which closely agree within the retention process (Fig. 1(d)). Unlike insertion, the fric-
5% of their root-mean-square value. tional force is upward during retention.
(3) Implemented through a commercial PHANTOM haptic device
The contact force (fn, fs lfn) is resolved into x and y components
[16], we demonstrate the effectiveness of the inverse model
so that solutions presented here can be compared against those
(as embedded algebraic solutions) for use in haptic render-
commonly used in snap-fit design for product assembly. From the
ing in design processes, where real-time virtual force feed-
equilibrium of forces in the x and y directions as illustrated in
back is essential.
Figs. 1(c) and 1(d)
    
2 Analytical Model 6fs cos c0  sin c0 fx
0 0 (2a)
Figure 1 illustrates the snap-fit assembly of a typical cantilever fn sin c cos c fy
hook (base thickness ho, width w, and length lt) with a wedge-
shaped end characterized by the height hb and angles (a, b), where 
the shaded cantilever indicates its initial state, and d is the beam 0 a0 a tan1 d=x; for x m ; t 
where c
deflection as the matching part contacts the wedge at x. As the b0 b  tan1 d=x; for x b ; m 
matching part advances (or retracts) for assembly (or disassem-
bly), the contact point slides along the front (or rear) surface of
the wedge as well as deflects the beam. In Eq. (2a), fs and fs refer to insertion and retention, respec-
Without loss of generality, the following geometries are consid- tively. Solving Eq. (2a), the assembly force fx and deflecting force
ered: x 0; t , a; b 0; p=2, and
8
< ho ; for x 0; b 
hx ho x  b tan b; for x b ; m ; b 0; p=2
:
ho t  x tan a; for x m ; t ; a 0; p=2
(1)

In Eq. (1), hx ho hb when a p=2 (m t ) or b p=2

(b m ). The following derivation focuses on assembly; the pro-
cess starts from x; y t ; ho =2. To offer intuitive insights to
facilitate design, we normalize the forces to (Ewho) and geometri-
cal dimensions to ho as follows:
   
Fx 1 fx x b hb t
; X ; L b ; Hb ; L t
Fy Ewho fy ho ho ho ho

The snap-fit assembly (moving from x t in the x direction)

consists of three processes: insertion, dwelling, and retention.
During insertion (Fig. 1(c)), the matching part moves up
along the inclined surface of the wedge (against a downward Fig. 2 Xi as a percentage of (X X )

121004-2 / Vol. 133, DECEMBER 2011 Transactions of the ASME

Downloaded 18 Dec 2011 to 222.205.33.68. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
 fy are related by Eq. (2b), where l is the friction coefficient Substituting Eq. (6) into Eq. (3b) leads to Eq. (7), the normalized
between the two sliding surfaces deflection (D d=ho) in Eq. (3b) becomes

8 D
0 1 X X1 X2 (7)
fx < tanc tan l Insertion sFx
l Dwelling (2b)  
fy : 1
tanc0  tan1 l Retention X1 6X Hx Lb (7a)
2
  x 2
2.1 Deflection Formulation. The assembly of a snap-fit can 1 ho dx
X2 6X Hx (7b)
be considered as a nonlinear beam deflection problem with the 2 b h3 x
movement of the matching part as an input (that leads to the 
changing contact position on the inclined surfaces of the cantile- 1; for x b ; m
s
ver hook). The beam deflection d can be expressed as 1; for x m ; t 
 (7c)
1 assembly when x m
d d d (3) s
1 disassembly:
where d and d are due to fy and fx, respectively, and the sub-
scripts and  indicate the deflections are in the y and y where h(x) is defined in Eq. (1) and Hx hx =ho . In Eq. (7), the
directions, respectively. Derived using Timoshenko beam theory sign s accounts for the direction of the induced moment, and at
(that takes into account shear deformation) with Castiglianos x m , it takes the direction of the matching part defined in Fig.
method, d and d are given by Eqs. (3a) and (3b) 1, positive for disassembly and negative for assembly.
2.2 Solution to Forward Model for Assembly. The normal-
@Ve ized beam deflection D d=ho in Eq. (3) can be compactly writ-
d (3a)
@fy ten as
1 1 @ DVe  
d  xh x (3b) X D=Fy X s fx =fy X (8)
2 2 @M

2.1.1 Deflection Due to Vertical Force Component. The In Eq. (8), X and X are defined in Eqs. (5) and (7), respectively.
strain energy due to fy is given by Eq. (4) The dimensionless mechanical impedances (X1, X2, X3) and
(X1, X2) for the geometry in Fig. 2 are given in the Appendix.
x x
fy x2 kfy2 The inverse model, which determined Fy in Eq. (8) for a given
Ve dx dx (4) deflection and force ratio fx=fy in Eq. (2b), can be solved in closed
0 2EI 0 2GA
form
where E and G are the Youngs modulus and shear modulus of the
beam material; A and I are the cross-sectional area and its moment D
Inverse model : Fy   (9a)
of the beam; and k is a dimensionless coefficient related with the X s fx =fy X
section shape (k 6=5 for rectangular cross-section). The normal-
ized deflection (D d=ho) is a function of Fy and can be The forward model which explicitly solves for the overall imped-
expressed in the form of mechanical impedance ance (X D=Fy ) must account for the force ratio fx=fy (which
depends on contact location and deflection and hook geometry)
and is rewritten from Eq. (8) as
D X 3
X Xi (5)     
Fy i1 Fy Fy
Forward model : g X2  gX  sX r X
where X1 4KL3b (5a) X X
x rX sgX 0 (9b)
12x2 dx 
X2 (5b)
3
b h x tanc bc Insertion and Dwelling
where g (10a)
x tanc  bc Retention
kE dx 
X3 (5c)
G b hx 1; Insertion and dwelling
r (10b)
  1; Retention
0:3 E 
K 1 2 (5d) 1; for x b ; m
Lb G s (10c)
1; for x m ; t 
During dwelling, the matching part simply slides along the tip.
In Eq. (10a),
2.1.2 Deflection Due to the Moment Induced by the Horizon-
tal Force. The strain energy induced by the component force fx bc tan1 l (10d)
which results in a bending moment M in the beam is given by 
a; for x m ; t 
Eq. (6) c (10e)
b; for x b ; m 
x
1 M2 The closed-form solution to Eq. (9b) is given by Eq. (11) where X
DVe dx (6)
2E 0 Ix must be real (4ac=b2  1)
 q
where b
X 16 1  ac2=b2 (11)
   2a
ho t  x tan a; for x m ; t 
M fx hx and hx
2 x  b tan b; for x b ; m  where a gFy =X (11a)

Journal of Mechanical Design DECEMBER 2011, Vol. 133 / 121004-3

Downloaded 18 Dec 2011 to 222.205.33.68. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
 Table 1 Characteristic dimensions of snap-fit 1
b gX  sX Fy =X r (11b)
c rX sgX (11c) a b Lt Lb Hb Lm l

25 deg 50 deg 7 5 0.67 5.56 0.1

2.3 Effects of Simplified Models on Assembly Force
Calculations. The overall deflection is contributed by component
impedances (X1 ; X2 ; X3 ) in Eqs. (5a)(5c) due to the vertical
force and (X1 ; X2 ) in Eqs. (7a) and (7b) due to horizontal
The induced moment due to fx is negative (or s 1) when
induced moment, where X16 and X26 are computed on the x 2 m ; t  during insertion and must be overcome when
uniform section and the hook shape of the cantilever beam, deflecting the beam. Both simplified models which neglect
respectively, and X3 accounts for the shear deformation. The this effect grossly underestimate the deflecting force. On the
magnitudes of these component impedances as a percentage of other hand, the induced moment is positive (or s 1) when
(X X ) are compared in Fig. 2 where the hook dimensions are x 2 b ; m , which tends to deflect the beam during the pro-
a 45 deg, b 90 deg, and Hb 1, and data were computed at cess of retention; as a result, both simplified models overesti-
X Lb 0:5 for insertion. The magnitudes of X1 and X1 , mate the deflecting force.
At the instant when the process changes from insertion to
which, respectively, increases and decreases with Lb , are generally
significantly higher than those of X2 ; X3 ; and X2 . These dwelling, the force ratio fx=fy suddenly increases in value to
straight-forward comparisons often lead designers to simplify l. Similarly, the deflecting force drops when transitioning
analyses by neglecting the hook geometry or approximate it as an from dwelling to retention. Both these two discontinuities,
EulerBernoulli beam which assumes no shear deformation insertion-to-dwelling and dwelling-to-retention, cannot be
(G!1), K 1, and X3 0 in Eq. (5). accounted for by either SM1 or SM2.
Two commonly used simplified models are discussed here. 2.4 Design Analyses for Disassembly. Disassembly of snap-
1. A common design practice is to treat the cantilever hook fits follows similar derivations for assembly except that the match-
as a uniform beam and neglect the bending moment M. ing part moves from x; y b ;  ho =2 in the x direction. For
The result is essentially the first term in Eq. (5a) with calculating the forward and inverse solutions in disassembly from
Lb ! Lt . Eqs. (11) and 9(a), Insertion, Dwelling, and Retention in Eqs. (2b)
and (10a) are replaced with Detachment, Dwelling, and
  Release which occur in sections x b ; m ; m ; m ; t  with
Simplified model 1 SM1 : X D=Fy  4KL3x (12)
slope angles c b; 0; a, respectively.
Figure 4 shows the normalized assembly force Fx involved in a
The hook geometry [hb, a, b] cannot be accounted for when typical cycle of assembly and disassembly of a snap-fit (dimen-
Eq. (12) is used. sions given in Table 1). When x 2 m ; t , the frictional force
2. A second approximation accounts for the hook geometry but opposes the motion of the matching part as it slides upward along
neglects the horizontally induced moment M. the sloping surface of the deflected hook during assembly but acts
in the opposite direction during disassembly. As derived in Eq.
Simplified model 2 SM2 : (2b), the force ratio fx =fy equals to tanc0 tan1 l during assem-
(13)
X D=Fy  X 4KL3b X2 X3 bly while it equals to tanc0  tan1 l during disassembly. Thus,
for the same slope (with angle c0 a0 ), the required insertion
This is a special case of Eq. (9b) when Fy =X  1 and g  0. force during assembly is larger than that required for release as
As in SM1, SM2 neglects the effects of friction and fx. shown on the left half of Fig. 4. Similar arguments can be made
on x 2 b ; m with slope angle b, where the required detaching
Figure 3 illustrates the effect of model approximation on the force is larger than retention force. Recall Eq. (7c) that s 1 for
deflecting force for assembling the snap-fit 1 with dimensions assembly and 1 for disassembly during dwelling; Fx lFy for
given in Table 1. Computed using SM1 and SM2 for Insertion, assembly is slightly higher than that for disassembly as given in
Dwelling, and Retention, the results are compared against the Eq. (8).
CM or Eq. (9a), which accounts for the hook geometry and bend-  
ing moment with shear deformation. Two observations in Fig. 3 (1) When b < bc tan1 l, the ratio fx =fy in Eq. (2b) is
are discussed. negative in retention. In practice, it implies that the

Fig. 4 Forces computed using CM for assembly and

Fig. 3 Comparing SM1 and SM2 against CM disassembly

121004-4 / Vol. 133, DECEMBER 2011 Transactions of the ASME

Downloaded 18 Dec 2011 to 222.205.33.68. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
 cantilever hook will never rebound to the original nonde-
formed position; the cantilever hook cannot be disas-
sembled. When l > cot b during insertion, the ratio fx =fy
in Eq. (2b) is negative, which means cantilever hook is too
steep to be disassembled.
(2) When b ! p=2 (and l  0) such that g ! 1 during
detachment, Eq. (9b) becomes independent of g and
reduces to

Fy D2 DX XX 1 (14)

Noting that Hx Hb at X Lm , the required force to

deflect the beam by D Hb is thus given by

Hb2

Fy X Lm

4KHb 6Hb 3L3m

The cantilever hook can only be unlocked with Fy. A com-

mon application is an automatic switch (such as a circuit Fig. 5 Insertion force versus displacement
breaker for mechanical equipment or power control), which
is normally in contact and breaks open to cut off current 3.2 Estimation of Friction Coefficient Using an
automatically when short circuit occurs. Outlet. Figure 6(a) demonstrates the use of an outlet (com-
3 Results and Discussions monly in household products) to determine the friction coefficient
between two material surfaces of a snap-fit. As compared in Fig.
Three sets of results are discussed here. The first set numeri- 6(b), three different outlet design configurations (DCs) are illus-
cally compares computed results against two different methods trated here as a means of experimental validation, each of which
for verifying the model validation. The second set illustrates the consists of a cold-drawn brass plug (E 0.93 GP and t 0.35)
use of the models for identifying the friction coefficient. The ex- and a fixed socket made of rolled phosphor bronze (E 1.13 GP
perimental results (obtained from three different design configura- and t 0.41). The extraction of the plug from the socket is essen-
tions (DCs) with surface materials) offer a means to validate the tially the dwelling process of a snap-fit, and thus follows Eq. (2b)
models. The third set demonstrates a unique application where the or fx=fy l which can be used to determine the friction coefficient
closed-form solutions are used for haptic rendering during design. between the two different material (socket and plug) surfaces.
Thus, from Eq. (8) with s 1
3.1 Model Verification. The accuracy of the model has been
verified by comparing results against two different methods: a X D dmax
l where Ew (16)
three-dimensional FEM implemented on ANSYS workbench [17] D=Fx  X Fx fx
and an approximate solution in Ref. [9]. The values of geometrical
and FEM parameters and material properties used in simulating In Eq. (16), the critical value of fx can be determined experimen-
the snap-fit 2 for these comparisons, which are exactly the same tally by gradually increasing the weight (mg) until the plug slips
as in Ref. [9], are given in Table 2. The results are compared in off the socket as shown in Fig. 6(a); X and X are given by Eqs.
Fig. 5, where we define the matching part displacement x0 so that (5) and (7), respectively, and the maximum deflection dmax can be
comparisons can be made on the same coordinate determined from the thickness of the plug. The experimentally
2 s 3 determined critical fx and the corresponding computed friction
2 coefficient are given in Table 3.
b b
x t  4
0
x d 5 2
(15) Since the three different outlet designs are made of the same
2 2 materials, they should theoretically have the same friction coeffi-
cient. The root-mean-square value of the three friction coefficients
Equation (8) has been derived with the assumption that h is small (0.123, 0.136, and 0.142) obtained experimentally from three dif-
and that the length of t  x0 approximately composes of b =2 and ferent outlet designs is 0.134. The maximum difference is within
q
5% of the root-mean-square value.
the hypotenuse x  b =22 d2 . Unlike the JK approximation
[9] which neglects the deflection due to the horizontal moment 3.3 Haptic Evaluation. It is desired that realistic force feed-
given in Eq. (6), the proposed method agrees well with the FEA as back can be felt virtually by designers when optimizing geometry.
compared in Fig. 5. This can be achieved by means of a haptic device (PHANTOM [16])
incorporating the sense of touch and control into the computer as
illustrated in Fig. 7.
Table 2 Geometrical parameters, material properties FEM of 3.3.1 Design Options. Performance of cantilever hook is
snap-fit 2 determined by various parameters. For the detachable snap-fits,
users prefer to assemble with least effort (or the minimum inser-
Geometry Materials Elements in ANSYS tion force jfxj) while allowing for ease of disassembly. This repre-
sents a design trade-off between the insertion and the retention
a, b(deg) 45,90 E(MPa) 2400 Beam: SOLID 186 angles for a specified hook length (t  b ) and offset hb. Thus,
lt (mm) 57.5 t 0.45 Contact: CONTA174, TARGE170
several combinations of different insertion and retention angles
lb (mm) 50 l 0 Number of elements: 7086
lm (mm) 50 with the same maximum offset are considered here and presented
ho (mm) 2.5 Substeps number to 1s : 102 in Table 4. As an illustration, the design objective here focuses to
hb (mm) 7.5 Mapped meshing method determine a preferred set of a and b.
w (mm) 5 Assembly feedback force is calibrated using a set of forces (5,
10, 15, and 20 N) to permit the designer to distinguish discrete

Journal of Mechanical Design DECEMBER 2011, Vol. 133 / 121004-5

Downloaded 18 Dec 2011 to 222.205.33.68. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
 Fig. 6 Outlet design configurations and friction coefficient; (a) determining l and (b)
three different outlet design configurations

Table 3 Experimental data for calculating friction coefficient

Type DC1 DC2 DC3

Socket (mm) b 5, m 6.9, b 60 deg, b 3.0, m 7.1, b 35.3 deg, b 5.4, m 9.6, b 58.8 deg,
hb 3.2, ho 0.5, w 7 hb 2.5, ho 0.6, w 6.8 hb 1.8, ho 0.6, w 8
Plug (mm) width 6.3, thickness 1.5
Deflection (mm) dmax 0.75 dmax 1.2 dmax 0.75
Critical fx (N) 160 g 1070 g 260 g
l 0.136 0.123 0.142

steps of forces for a specified set of design values (including mate- snap-fit design (geometry and material) for a given contact point
rial and geometry parameters). For the detachable snap-fit, the and force. Haptic rendering, which can enhance the snap-fit design
assembly force fa < 10 N and disassembly force fd > 10 N. With for assembly, requires the feedback of virtual forces in real time
these constraints, nine design options listed in the first four col- and thus the solutions to the inverse model which computes the
umns in Table 4 are chosen as inputs to the haptic algorithm. force from the deflection and contact position. Once the location
X (and hence the deflection D) of the matching part is given, the
3.3.2 Inverse Model for Haptic Rendering. For real-time deflecting force fy can be computed from Eq. (8), and the corre-
applications, rapid haptic feedback is achieved using embedded sponding assembly=disassembly force fx can be obtained from
algebraic solutions given in Sec. II. The solutions to the forward Eq. (2b).
model (9b), however, compute the mechanical impedances for
3.3.3 Haptic Interface and Human Force. The Haptic Device
API [16] is employed here to enable the designer to obtain feed-
back force information in real time directly with the assistance of
SENSABLE OPENHAPTICS. In Fig. 7, the position of PHANTOM stylus as
carrier for moving object is detected in high frequency (1 kHz).
Users evaluate the reaction force in virtual environment. The feed-
back forces for the nine different cantilever design options have
been recorded (Fig. 8) while experienced by users with the assis-
tance of haptic device. In Fig. 8, the horizontal and vertical axes
represent the displacement of the matching part and the feedback
force, respectively.
Nine individuals participated as design evaluators with the aid
of the haptic device. Each evaluator experienced a haptic feel-
ing of the insertion force, retention force, and the abrupt change
while the insertion angle was stepped up from 18.5 until 45. The
nine different labeled design options (Table 4) were randomized
before each blindfolded evaluator conducted the sequential evalu-
ation of the nine design configurations to determine the preferred
option; only one trail was allowed for each evaluator. Note that
beta is dependent on alpha for a specified lb, hb, and lt. When beta
equals to 90 deg, 76.6 deg, 69.2 deg, 63 deg, 57.8 deg, and 53.5
deg, the corresponding alpha values are 18.5 deg, 20 deg, 21 deg,
22 deg, 23 deg, and 24 deg. As shown in Fig. 8, there are signifi-
Fig. 7 Haptic evaluation procedure cant differences in insertion and detaching forces among some

121004-6 / Vol. 133, DECEMBER 2011 Transactions of the ASME

Downloaded 18 Dec 2011 to 222.205.33.68. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
 simplified models may be desired to facilitate design optimization
and real-time control. The relation between feedback force
and deflection has been deduced illustrating a complete cycle of
snap-fit assembly and disassembly; the results offer some insight-
ful findings by comparing several commonly used simplified
models.
The model presented has been verified by comparing against
results obtained by ANSYS FEA workbench and published approxi-
mations and applied to two applications. The first illustrates its
use to determine the friction coefficient between two contact
surfaces. For this, the friction coefficient involved in three differ-
ent designs (with same materials) was compared which closely
agree within 5% of their root-mean-square value. The second
demonstrates the effectiveness of the inverse model (as embedded
algebraic solutions) for use in haptic rendering in design proc-
esses, where real-time virtual force feedback is essential. Imple-
mented through a commercial PHANTOM haptic device, nine
individuals participated in evaluating nine design options with dif-
Fig. 8 Haptic feedback curve ferent parameters settings. The experiment showed that 78% of
them chose the optimal theoretical solution by feeling the feed-
(a, b) combinations. Although there appears little differences back force. Through a relatively accurate virtual force feedback, it
among the insertion forces for the designs with insertion angles of is expected that the confidence of assembly robustness can be
18.5 deg, 20 deg, 21 deg, 22 deg, 23 deg, and 24 deg, the detaching enhanced in industrial settings.
forces differ significantly (as shown in Fig. 8) from each other.
The results are given in the last two columns in Table 4, which Acknowledgment
aim at satisfying two constraints: fa < 10 N and fd > 10 N. As
seen in Fig. 8, options IVII satisfy constraint fa < 10 N, while The paper was supported by Major State Basic Research Devel-
options IV satisfy constraint fd > 10 N. In other words, only opment Program of China (No.2011CB706506) and the National
options IV satisfy both these constraints, and based on the rule of Natural Science Foundation of China (No.50775201).
least effort, option V is chosen as the optimal design. Two of the
users chose option IV; it is worth noting that the felt force differ- Nomenclature
ence between option IV and option V is less than 5 N. Symbols
A beam cross-sectional area
E Youngs modulus
4 Conclusions f force
An analytical model for design cantilever hook has been pre- F normalized f
sented, which has potential for applications where real-time haptic G shear modulus
evaluation through feedback force of a humancomputer interfac- h(x) cantilever hook thickness along x
ing mechanism is essential or represents an advantage. This rela- hb height of the wedge-shaped hook
tively complete model, which takes into account the hook-shape Hb normalized hb
geometry and the effect of shear deformation which cannot be ho base thickness of the cantilever hook
neglected for thick elements, provides a basis for justifying I moment of inertia
assumptions made to neglect certain factors in applications where k dimensionless section shape coefficient
lb beam length
Lb normalized lb
Table 4 Design options and users selection lm distance between hook tip and beam root
Lm normalized lm
lt cantilever hook length
Lt normalized lt
M moment
s direction of the induced moment
V strain energy
w cantilever hook width
x,y contact point coordinates defined at beam root
X, Y normalized x and y

Design options Geometry User Evaluation Greek Symbols

a; b initial wedge-shaped hook angles
a b (mm) choice conclusion a0 ; b0 deflected wedge-shaped hook angles
bc friction angle
I 18.5 90 V d beam deflection
II 20 76.6 lt 35 V 78% users D normalized d
III 21 69.2 lb 25 V select option V c, c0 initial and deflected angle defined in Eq. (2a)
IV 22 63.0 VI as preferred g global equivalent friction coefficient
V 23 57.8 V option, while l local friction coefficient
VI 24 53.5 w 10 V the remainder
VII 30 38.6 ho 5 VI selects
h beam neutral axis deflection angle
VII 35 32.7 V option VI. r the direction of the matching part moving
IX 45 26.7 V t passion ratio
X mechanical impedance

Journal of Mechanical Design DECEMBER 2011, Vol. 133 / 121004-7

Downloaded 18 Dec 2011 to 222.205.33.68. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
 x2
Subscripts dx 1
x, y along x or y direction lnja bxjjxx21
x1a bx b
n normal to the slope surface " #
x2
x2

s parallel along the slope surface dx 1

positive y direction under deflection

x1 a bx
3
2ba bx2
x
 negative y direction under deflection 1

a assembly
d disassembly For completeness, the results are summarized in Eqs. (A1)(A5)
e strain where qho 1=1 Hb , qx 1=1 Hx , and qr qho =
max maximum value qx:
  
0:3 E
Appendix X1 4 1 2 L3b (A1)
Lb G
The integrals in Eqs. (5b), (5c), and (7b) can be analytically
solved by noting that
x2 " #
x2
x2 dx 1 2a a2

ln j a bx j 

x1 a bx
3 b3 a bx 2a bx2

x1

8    
>
> 6 2 ln qho  4Hb qho 1  Lb tan b 6 2 ln qr 41 Lt tan aqx  qho  Insertionand
< 2  2
tan3 b  Hb 2 Hb 1  Lb tan b q2 ho  tan3 a 1 Lt tan a q2 x  q2 ho dwelling
X2
>
> 6 2 ln qx  4H x qx 1  Lb tan b
: 3 2 Retention
tan b Hx 2 Hx 1  Lb tan b q2 x
(A2)

8  
>
> kE ln qr ln qho X1 3X2Hx 1Lb ; Insertion; dwelling; and retention (A4)
<  Insertion and dwelling
G tan a tan b
X3
>
> kE ln qx
: Retention
G tan b
(A3)

8  2 
>
> 3X q x  q2 ho q2 ho  1
< 2Hx 1  Insertion and dwelling
X2 2 tan a tan b (A5)
>
> 3X  
: 2Hx 1 1  q2 x ; Retention
2 tan b

References [10] Rusli, L., Luscher, A., and Sommerich, C., 2010, Force and Tactile Feedback
[1] Bonenberger, P. R., 2000, The First Snap-Fit Handbook: Creating Attachments in Preloaded Cantilever Snap-Fits Under Manual Assembly, Int. J. Ind. Ergon.,
for Plastics Parts, Hanser Gardner Publications, Cincinnati, OH. 40(6), pp. 618628.
[2] AlliedSignal Corporation, 1997, Modulus Snap-Fit Design Manual, Allied-Sig- [11] Rose, D., Bidmon, K., and Ertl, T., 2004, Intuitive and Interactive Modifica-
nal Plastics, Morristown, NJ. tion of Large Finite Element Models, IEEE Visualization 2004 Conference,
[3] Bayer Corporation, 1996, Snap-Fit Joints for Plastics, A Design Guide, Poly- Austin, TX, pp. 361368.
mer Division, Pittsburgh, PA. [12] Yang, Z. Y., Lian, L. L., and Chen, Y. H., 2005, Haptic Function Evalu-
[4] Luscher, A. F., 1996, Part Nesting as a Plastic Snapfit Attachment Strategy, ation of Multi-Material Part Design, Comput.-Aided Des., 37, pp.
Proceedings of the 1996 54th Annual Technical Conference Part 1 (of 3), India- 727736.
napolis, IN, pp. 13021306. [13] Feng, H., 2007, Robot-Assisted Suspension Laryngoscopy Virtual Simulation
[5] Genc, S., Messler, R. W., and Gabriele, G. A., 2000, A Method for Attachment System, Ph.D. thesis, Tianjin University, Tianjin, China.
Design Concept Development in Integral Snap-Fit Assemblies, ASME J. [14] Gao, Z., and Lecuyer, A., 2009, Path-Planning and Manipulation of Nanotubes
Mech. Des., 122(3), pp. 257264. Using Visual and Haptic Guidance, IEEE International Conference on Virtual
[6] Suri, G., and Luscher, A. F., 2000, Structural Abstraction in Snap-Fit Analy- Environments, Human-Computer Interfaces and Measurements Systems, Hong
sis, ASME J. Mech. Des., 122(4), pp. 395402. Kong, China.
[7] Suri, G., and Luscher, A. F., 2006, Development of Analytical Model of Canti- [15] Wang, L., Wang, H. G., and Pen, Q. S., 2009, Calculation and Tactile Percep-
lever Hook Performance, ASME J. Mech. Des., 128(2), pp. 479493. tion for Molecular Force Field With CHARMM-Based Field, J. Comput.
[8] Lan, C., and Lee, K. M., 2008, An Analytical Contact Model for Design of Appl., 21(7), pp. 886893.
Compliant Fingers, ASME J. Mech. Des., 130(1), 011008. [16] OPENHAPTICSV R Toolkit version 3.0 Programmers Guide.
[9] Jorabchi, K., and Suresh, K., 2009, Nonlinear Algebraic Reduction for Snap- [17] ANSYS, Inc., www.Ansys.com.
Fit Simulation, ASME J. Mech. Des., 131(1), 061004.

121004-8 / Vol. 133, DECEMBER 2011 Transactions of the ASME

Downloaded 18 Dec 2011 to 222.205.33.68. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm