Escolar Documentos
Profissional Documentos
Cultura Documentos
DOI 10.1007/s00170-008-1790-0
ORIGINAL ARTICLE
Received: 3 June 2008 / Accepted: 2 October 2008 / Published online: 31 October 2008
# Springer-Verlag London Limited 2008
Abstract Mechanical products are usually made by assembling many parts. The dimensional and geometrical
variations of each part have to be limited by tolerances
able to ensure both a standardized production and a certain
level of quality, which is defined by satisfying functional
requirements. The appropriate allocation of tolerances
among the different parts of an assembly is the fundamental
tool to ensure assemblies that work rightly at lower costs.
Therefore, there is a strong need to develop a tolerance
analysis to satisfy the requirements of the assembly by the
tolerances imposed on the single parts. This tool has to be
based on a mathematical model able to evaluate the
cumulative effect of the single tolerances. Actually, there
are some different models used or proposed by the literature
to make the tolerance analysis of an assembly, but none of
them is completely and univocally accepted. Some authors
focus their attention on the solution of single problems
found in these models or in their practical application in
computer-aided tolerancing systems. But none of them has
done an objective and complete comparison among them,
analyzing the advantages and the weakness and furnishing
a criterion for their choice and application. This paper
briefly introduces two of the main models for tolerance
analysis, the vector loop and the matrix. In this paper, these
models are briefly described and then compared showing
their analogies and differences.
1 Introduction
As technology increases and performance requirements
continually tighten, the cost and required precision of
mechanical assemblies increase as well. Then, there is a
strong need for industries to produce high-precision
assemblies at lower costs. Therefore, there is a strong need
to use tolerance analysis to predict the effects of the
tolerances that have been assigned to the components of an
assembly on the functional requirements of the assembly
itself. The aim of the tolerance analysis is to study the
accumulation of dimensional and/or geometric variations
resulting from a stack of dimensions and tolerances. The
results of the analysis are meaningfully conditioned by the
adopted mathematical model. Some are the models proposed by the literature to carry out a tolerance analysis of
an assembly, but they still appear not adequate under many
aspects: the schematization of the form deviations, the
schematization of the joints with clearance between the
parts, the solution of complex stack-up functions due to
the network joints among the components, and so on.
Moreover, there does not exist in the literature a paper that
compares the different analytical methods on the basis of a
case study that underlines in a clear way all the advantages
and the weakness. In the literature, some studies compare
the models for tolerance analysis by dealing with their
general features [1, 2]. Other studies compare the main
computer-aided tolerancing softwares that implement some
of the models of the tolerance analysis [3, 4]; but these
studies focus the attention on the general features. However,
a complete comparison of the models proposed to solve the
1107
1108
1
40
0
gi
cos fi
the following shape: Ri 4 sin fi
2
3
0
sin fi
cos fi
0
0
0 5 and Ti
1
0 Li
1 0 5 where i is the angle between the vectors at
0 1
du B1 A dx B1 F da
dg C dx D du G da
where Sx C D B1 A and Sa G D B1 F
are the sensitivity matrices. When the sensitivity matrices
jSxik txk j
k
X
l
jSail tal j
gi
hX
Sxik txk 2
k
Sail tal 2
l
i1=2
1109
mR
n h
X
i
P1
R!Ri Di I mRi
3 Models comparison
3.1 Case study
To compare the two models previously described, the case
study shown in Fig. 1 has been used. It is constituted by a
box containing two circles. The aim of the tolerance
analysis is the measurement of the variation of the gap g
between the second circle and the top side of the box (g)
as a function of the tolerances applied to the components.
The first analysis has considered only the dimensional
tolerances that are shown in Fig. 2. The envelope principle
has been applied, i.e., rule #1 of the American Society of
Mechanical Engineers (ASME) standards. Then, the tolerance
analysis has considered the geometrical tolerances too, as
10
11
12
i1
1110
G
g
H
x2 = 80 0.50
O2
x4
D
x4
C
u4
x4
x3 = 20 0.05
x3
O1
x4 = 20 0.05
x3
x3
u2
E
A
u1
u3
x1 = 50 0.20
1111
Table 1 Elements of R and T matrices of the loops (dimensional
tolerances only)
Circle 1
L1
L2
Box
Circle 2
L3
g
13
gWC
14
C
u4
1
2
3
4
5
0
90
7 13
90
90
u1
x3
x3
u2
1
2
3
4
5
6
7
8
gStat
Nr.
0
90
90
7 24
0
7 26
90
90
x1
u4
x4
x4
x3
x3
u2
1
2
3
4
5
6
7
8
0
90
90
7 34
0
90
90
90
x1
u4
x4
x4
g
u3
x2
hX
Sxik txk 2
i1=2
u3
15
u1
Loop 3
x4
x3
Nr.
x3
u2
x3
O1
x4
x4
Loop 2
Nr.
H
O2
Loop 1
1112
//
G 0.10
0.10
0.20
H
O2
0.05
x2 = 80 0.50
//
0.05
x4
D
x4
C
u4
x4
x3 = 20 0.05
x3
O1
G
g
x4 = 20 0.05
x3
x3
u2
E
A
u1
0.10
u3
x1 = 50 0.20
16
gWC
X
j Si j x i
X
Sj aj
1:0340 1:03 mm
17
Table 2 Elements of R and T matrices of the loops (dimensional and geometrical tolerances)
Loop 1
Loop 2
Nr.
1
2
3
4
5
6
7
8
9
0
90
0
0
7 13
0
0
90
90
u1
a1
a4
x3
x3
a5
a2
u2
0 0:05
0 0:025
0 0:025
0 0:05
Loop 3
Nr.
Nr.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
0
90
90
0
0
7 24
0
0
0
7 26
0
0
90
90
x1
u4
a3
a8
x4
x4
a7
a6
x3
x3
a5
a2
u2
1
2
3
4
5
6
7
8
9
0 0:1
0 0:025
0 0:025
0 0:025
0 0:025
0 0:05
0
90
90
0
0
7 34
0
0
0
x1
u4
a3 0 0:1
a8 0 0:025
x4
x4
a9 0 0:025
G
a10 0 0:05
1113
18
1114
19
20
1115
22
23
24
1116
sum of the four contributors due to the DRFs R5, R2, R6, and
R4:
yE yE;R5 yE;R2 yE;R6 yE;R4
25
gWC 0:69 mm
g yF yE yF
yE;R5 yE;R2 yE;R4 yE;R6
26
29
g u3 5 sin g 3 v5 20 sin g 2
38:73 1 cos g 2 v6 30 sin g 4
27
38:73 1 cos g 4
Equation 27 has to be maximized under the following
constraints due to the required tolerances (the details of the
mathematical steps are reported in Appendix):
0:50 u3 25 sin g 3 0:50
0:50 u3 25 sin g 3 0:50
0:10 50 sin g 3 0:10
0:05 v5 0:05
0:05 u5 0:05
0:05 0:25 u5 0:97 v5 0:05
0:20 u2 40 sin g 2 0:20
0:20 u2 40 sin g 2 0:20
0:20 80 sin g 2 0:20
0:05 u6 0:05
0:05 v6 0:05
0:05 0:2500 u6 0:9682 v6 0:05
0:10 80 sin g 4 0:10
28
4 Comparison
Table 4 shows the results obtained by the application of the
two considered models to the same case study. The second
column shows the values of the gap range (g) obtained by
means of the exact worst-case approach. The third and
fourth columns report the results obtained by the vector
loop and the matrix models.
The vector loop model has two advantages as regards to
the matrix model: it allows us to model form tolerances and
it may be solved by means of a statistical approach. The
vector loop model gives better results than matrix does
since the results are nearer to the exact solution than the
matrix ones.
It is possible to see that the matrix model gives always
an underestimate, while the vector loop model underestimates if only the dimensional tolerances are considered
and overestimates if both dimensional and geometrical
tolerances are taken into account. This is due to the fact that
the vector loop model sums the assigned tolerances by
1117
Worst case
Statistical
Worst case
Statistical
Exact solution
Vector loop
Matrix
0.89
0.91
0.78 (12%)
0.52
1.03 (+13%)
0.54
0.70 (21%)
0.69 (24%)
Appendix
Case study solution by vector loop model
with dimensional tolerances only
As concerning the first loop, Eq. 1 becomes:
R1 T1 R2 T2 R3 T3 R4 T4 Rf I
30
31
32
33
5 Conclusions
As concerning the third loop:
This paper firstly makes a brief review of two state-of-theart tolerance analysis models, the vector loop and the
matrix. Then, the two models are compared, in order to
highlight the advantages and the weakness of each model,
based on the experimental results and available information
from the literature.
The application of the models on the same case study
and the experimental results show how both the vector loop
and the matrix models have their advantages and the
weakness that are investigated and explained.
Further researches includes the definition of a new and
original model able to overcome the limits highlighted in this
work.
R1 T1 R2 T2 R3 T3 R4 T4 R5
T5 R6 T6 R7 T7 Rf G
34
that gives:
g x 2 u4 x 4
35
36
that gives:
du B1 A dx Su dx
37
1118
41
As concerning the third loop:
R1 T1 R2 T2 R3 T3 R4 T4 R5
42
T5 R6 T6 . . . R11 T11 Rf G
that gives:
38
g x2 a10 u4 x4 a9
43
Sud dx Sua da
du B1 A dx B1 C da
40
44
Sua
0
61
6
61
6
60
6
40
0
3
0
0
1
0
6
7
0
0
1
0
6
7
6
0:2582 0 2:2910
1:2910 7
7
Sud 6
6
7
0
0
0
0
6
7
4 0:0258 0 0:0323
0:0323 5
0:0258 0 0:0323 0:0323
1
0
0
1
0
0
0
0
1
0
0
0
0:2582
0:2582 1 0:2582
1:0328
1:0328
0
0
0
0
0
0
0:0258
0:0258 0 0:0258
0:0064
0:0064
0:0258 0:0258 0 0:0258 0:0064 0:0064
0
0
0:2582
0
0:0258
0:0258
0
0
0
0
0
0
3
0
07
7
07
7
07
7
05
0
cg cb
6 sg cb
6
D3 4
sb
0
sg
cg
0
0
cg sb
sg sb
cb
0
3 2
cg 3
u
6
07
7 6 sg 3
05 4 0
1
0
sg 3
cg 3
0
0
3
0 u3
0 07
7
1 05
0 1
45
1119
PR>R3
ca
6 sa
6
4 0
0
sa
ca
0
0
0
0
1
0
x
0
6 1
y 7
76
0 5 4 0
1
0
1
0
0
0
0
0
1
0
25
80 7
7
0 5
1
46
61
6
FF00 R P1
R>R3 D3 I FR3 6
40
2
6
6
6
4
1 cg 3
sg 3
sg 3
1 cg 3
0
0
0
0
0
0
1
80
25 7
7
7
0 5
0 0 0
1
3 2 3 2
3
0
5 1 cg 3
7
6 7 6
07
7 6 5 7 6 u3 5 sg 3 7
76 76
7
5
0 5 405 4
0
0
0
1
47
Therefore:
48
1 cg 3
sg 3
u3
sg 3
0
1 cg 3
0
0
0
0
0
0
0
0
3 2 3
1
0
6 25 7 6 0 7
6 7 6 7
6 7 6 7 u3 25 sg 3
4 0 5 405
6
6
D3 I HR3 x 6
4
2
3
u5
v5 7
7
05
1
3
7
7
7
5
52
0
1
0
20
20 7
7
7
0 5
1
Therefore:
yE;R5 v5
56
0
60
6
D5 I AR5 r 4
0
0
50
55
0
0
0
0
0
0
0
0
3
3 2
3 2
0
0
u5
7
7 6
6
v5 7
7 6 20 7 6 1 7 v5
05 4 0 5 4 0 5
0
1
0
53
54
u3
yF u3 5 sing 3
58
Substituting Eq. 58 in Eq. 57 the constraints on A point
becomes:
0:05 v5 0:05
59
60
61
1120
cg cb
6 sg cb
6
D2 4
sb
0
sg
cg
0
0
cg sb
sg sb
cb
0
u
cg 2
6
07
7 6 sg 2
05 4 0
0
1
that gives:
2
1 cg 2
sg 2
6
sg 2
1 cg 2
6
D2 I HR2 x 6
4
0
0
sg 2
cg 2
0
0
0
0
0
3
u2
07
7
7
05
0
0
0
2
3 2 3
1
0
6 40 7 6 0 7
6
7 6 7
6
7 6 7 u2 40 sg 2
4 0 5 405
0 u2
0 07
7
1 05
0 1
62
67
The homogeneous transformation matrix to pass from
the DRF R to the DRF R2 [PR->R2] is given by:
2
PR>R2
ca
6 sa
6
4 0
0
sa
ca
0
0
3 2
0 x
1
6 0
0 y 7
76
1 0 5 4 0
0 1
0
0
1
0
0
0
0
1
0
3
50
40 7
7
0 5
1
63
3
1 0 0 50
6 0 1 0 40 7
6
7
EE0 R P1
7
R>R2 D2 I ER2 6
4 0
0 1 0 5
0
0 0 1
2
3 2
3
sg 2
0 u2
1 cg 2
20
6
6
7
sg 2
1 cg 2 0 0 7
6
7 6 38:73 7
6
76
7
4
0
0
0 05 4 0 5
0
0
0 0
1
2
3
20 1 cg 2 38:73 sg 2 u2
6 20 sg 38:73 1 cg 7
6
7
2
2
6
7
4
5
0
68
69
PR>R6
ca
6 sa
6
4 0
0
3 2
0 x
1
60
0 y 7
76
1 0 5 40
0 1
0
sa
ca
0
0
0
1
0
0
0
0
1
0
3
30
58:73 7
7
0 5
1
71
64
Therefore:
yE;R2 20 sing 2 38:73 1 cosg 2
65
66
1 0
60 1
6
D
EE0 R P1
I
E
6
6
R>R6
R6
40 0
60
6
6
40
0
0
0
0
0 0
3 2 3 2 3
u6
0
u6
6 7 6 7
v6 7
7 6 20 7 6 v6 7
76 76 7
05 4 0 5 405
0
0
1
0
30
58:73 7
7
7
0 5
1
72
1121
Therefore:
yE;R6 v6
73
74
that gives:
2
0
60
D6 I DR6 x 6
40
0
0
0
0
0
0
0
0
0
3 2 3 2 3
1
20
u6
6 0 7 607
v6 7
7 6 7 6 7 u6
0 5 4 0 5 405
0
1
0
82
75
83
84
0:05 u6 0:05
76
77
0
60
6
D6 I CR6 r 6
40
0
0
3
3 2
5
0 u6
6
7
0 v6 7
7 6 19:37 7
7
76
0 05 4 0 5
0
1
0 0 0 0
2
3
0:25
6 0:97 7
6
7
6
7 0:25 u6 0:97 v6
4 0 5
0
79
80
81
subject to the constraints of Eq. 51, 52, 5961, 68, 69, 76,
77, and 80.
HH'R3 II'R3 x
D3 I HR3 IR3 x to
85
86
87
88
89
0:05 u5 0:05
90
91
1122
92
93
94
96
EE0 R P1
R>R4 D4 I ER4
2
1 cg 4
6
sg 4
6
6
4
0
1
60
6
6
40
0
sg 4
1 cg 4
0
0
0
1
0
0
0 0
1
3
3 2
30
0
6
7
07
7 6 38:73 7
7
76
05 4 0 5
0
0
0
0
0 0
2
3
30 1 cg 4 38:73 sg 4
6 30 sg 38:73 1 cg 7
6
4
4 7
6
7
4
5
0
0
104
97
98
99
0:05 v6 0:05
100
Therefore:
yE;R4 30 sing 4 38:73 1 cosg 4
101
cg cb
6 sg cb
D4 6
4 sb
0
sg
cg
0
0
cg sb
sg sb
cb
0
cg 4
u
6 sg 4
07
76
05 4 0
1
0
sg 4
cg 4
0
0
0
0
1
0
0
07
7
05
1
102
105
3
0
40 7
7
7
0 5
6
6
D4 I OR4 IR4 x 6
4
sg 4
sg 4
1 cg 4
07
7
7
05
0
0
3 2 3
1
6 80 7 6 0 7
6
7 6 7
6
7 6 7 80 sg 4
4 0 5 405
2
1 cg 4
107
Substituting Eq. 107 in Eq. 106, the constraints on points
O and I become:
108
1123
Substituting Eqs. 82, 88, 92, 98, and 105 in Eq. 49 is:
g u3 5 sing 3 v5 20 sing 2 38:73
1 cosg 2 v6 30 sing 4 38:73
1 cosg 4
109
References
1. Shen Z, Ameta G, Shah JJ, Davidson JK (2005) A comparative
study of tolerance analysis methods. J Comput Inf Sci Eng 5
(3):247256 doi:10.1115/1.1979509
2. Hong YS, Chang TC (2002) A comprehensive review of
tolerancing research. Int J Prod Res 40(11):24252459
doi:10.1080/00207540210128242
3. Salomons OW, van Houten FJAM, Kals HJJ (1998) Current status
of CAT systems. In: ElMaraghy HA (ed) Geometric design
tolerancing: theories, standards and applications. Chapman &
Hall, London, pp 438452
4. Prisco U, Giorleo G (2002) Overview of current CAT systems.
Integr Comput Aided Eng 9(4):373397
5. Standard ASME (1994) Dimensioning and tolerancing. ASME
Y14.5M-1994, American Society of Mechanical Engineers, New York
6. ISO Standard (1985) ISO 8015: fundamental tolerancing principle
7. Chase KW. Tolerance Analysis of 2-D and 3-D Assemblies
(automated method). http://adcats.et.byu.edu/home.php
8. Chase KW, Gao J, Magleby SP (1995) General 2-D tolerance
analysis of mechanical assemblies with small kinematic adjustments. J Des Manuf 5(4):263274