Académique Documents
Professionnel Documents
Culture Documents
I. INTRODUCTION
xi =
z
n
F
i =1
Fi
(1)
#132
In a continuous cost-tolerance model an algebraic
relationship is established between tolerance and cost,
typically using empirical data through regression analysis. A
typical cost-tolerance relationship is illustrated in Fig. 1.
Various functions have been proposed to represent a costtolerance relationship and various solution methods have been
applied (Table I). Comparisons of some of the cost models and
solution techniques used can be found in ElMaraghy and
ElMaraghy [24]. The main drawback with these strategies is
that the required cost data is not available, especially at the
early design stage.
Ostwald and Huang [19] proposed a tolerance allocation
strategy that minimizes cost with process selection based on a
zero-one algorithm presented by Balas [25]. Lee and Woo [22]
proposed another allocation strategy based on minimizing cost
with process selection using a selection method called the
branch and bound method. Chase and Greenwood [23]
proposed similar allocation strategies, but used different search
algorithms for process selection viz. exhaustive search method,
univariate search method, sequential quadratic programming
algorithm and others. Comparisons of all the allocation
strategies by minimizing costs with process selection
mentioned above can be found in [23]. These strategies use
discrete cost-tolerance models which are promising;
nevertheless the question of the availability of reliable data
remains.
Taguchi introduced the concept of the quality loss of a
product [26]. According to this concept, all critical parameters
(including dimensions) of a product should be at their target
values, which will ensure the products best performance.
Several researchers, such as Krishhnaswami and Mayne [27],
Sderberg [28,29], Choi and Park [30] Kapur [31], Fathi, et al
[32] and Jeang [33], have based their tolerance allocation on
minimizing quality loss and/or productivity loss, which is the
sum of quality loss and production cost. However, these
strategies require production cost data as well as
replacement/repair cost data, which are often not available at
the early design stage.
In addition, there are several other tolerance allocation
strategies that use relatively new solution techniques, such as:
expert system [34], fuzzy logic [35], neural network [36], genetic
algorithm [37,38], and experimental design [39].
2
TABLE I
COST-TOLERANCE MODEL EXAMPLES
Model Name
Cost Model
Linear
A - Bx
Exponential
Be
-mx
-mx
Be
Solution Method
Reference
Linear programming
Graphical
Peters [6]
Lagrange multiplier
Lagrange multiplier
Speckhart [15]
Lagrange multiplier
Spootts [16]
Exponential
Be
Reciprocal power
A + B/x
Empirical data
Discrete points
Lagrange multiplier
Zero-one programming
Nonlinear programming
Reciprocal
A + B/x
Nonliner programming
Parkinson [21]
Reciprocal
A + B/x
Lagrange multiplier
Piecewise linear
Ai - Bixi
Linear programming
Bjrke [7]
Empirical data
Discrete points
Empirical data
Discrete points
Lagrange multiplier
#132
Formulate
Problem
Yes
Enter values of
non-negotiable
dimensions
No
Modify processes
through appication of MRBT
Modify design or
Modify assembly strategy
Guided
Redesign
Evaluate
Alternatives
RT = 0
Final Results
STOP
IT 16
5
(2)
Cp
RT < 0
RT = 0
RT > 0
(3)
Generate
Alternatives
RT < 0
Non-negotable
dimensions?
RT > 0
Want to
utilize RT?
No
#132
RT = z xi
(4)
i =1
(5)
RT
(6)
100 %
xi
The required percentage change for each tolerance shows how
much improvement is necessary to fulfill a target tolerance
requirement if only one process is altered. Nevertheless, these
values should be considered as indicative only. The CE team
selects those options that have the greatest effect and seem
feasible. The options must be translated into specific proposals.
For example, if the problem can be solved by a slight
improvement, then an option will be to remove the uncertainties
of that process so that the safety factor can be reduced. It is
estimated that if the IT grade is reduced by one, then the
tolerance is reduced by 37 percent. By using this information a
new process can be proposed.
When proposing a process alternative, all avenues for
achieving the required finish tolerance should be considered,
including whether to replace the finishing operation only, to add
a new finishing operation, to change the whole process plan, or
( RPC)i
Tu = t m + t h + t cu = t m + t h + t c
tm
T
(7)
#132
the tool change time per unit; tc is the tool change time; and T
is the average tool life.
It is worth noting that the total time required to produce a
unit (min/unit) is the reciprocal of the production rate
(units/min); thus, the minimization of unit production time will
lead to the maximization of the production rate.
If desired, the CE team members can calculate each time
factor and make a decision on that basis. However, this is time
consuming and the team members are interested in a
comparative assessment (i.e., which option will least increase
the total production time per unit). Rather than an absolute
value, a decision matrix (Table II) can be formed for this
purpose and, taking one option as a datum (the existing
option), the effect of other options can be assessed. This
method is suitable for combining the judgments of team
members for decision making. The details of the decision
matrix and its use can be found in [50]. After deciding which
process is to be changed the CE team can recalculate the
residual tolerance and evaluate the outcome. The process will
be repeated until an acceptable solution is found.
After the allocation of tolerances according to the proposed
manufacturing process capabilities, if there is slack (i.e.,
residual tolerance is positive), then a decision can be made on
the basis of the magnitude of the residual tolerance. It is worth
noting that increasing the selected tolerances beyond the
achievable process capabilities may prove to be counterproductive as it generally does not reduce the cost. Moreover,
it might produce misleading specifications for the part, and if
the part is used in another assembly, problems might arise.
Therefore, the recommended option will be to change the
process and, for example, select a process with less precision
(higher IT grade). The possible change options can be
generated from the required percentage change of each
selected tolerance.
TABLE II
DECISION MATRIX FOR COMPARING ALTERNATIVE OPTIONS.
Criteria
Weighting
IV. AN APPLICATION
Purchased Parts:
Retaining Ring, (A a) = ( 1.283 0.019) mm
Bearing,
(C c) = (12.936 0.032) mm
Bearing,
(G g) = (12.936 0.032) mm
The tolerance allocation procedure for the above example is
demonstrated in Tables III-V. This assembly contains three
purchased parts. Their dimensions are non-negotiable and are
deducted from the target tolerance value (Table III). Following
this, a turning operation is proposed for the manufacture of all the
remaining features. The process capability database is consulted
and the IT grade range for turning operations under average
conditions is recorded. The difficulty level of the manufacture of
each feature is considered and the IT grades for each dimension
and process under real conditions are determined. Then, the
achievable process capability tolerances are calculated using
Equation 2. The process capability tolerances are enlarged by
taking into account the uncertainties of each process.
After allocating the tolerances on the basis of enlarged process
capability, it is found that the residual tolerance is negative (0.015 mm). This indicates that some corrective action is
necessary. From the required IT change calculations, it can be
seen that to achieve the target values for the functional equation
concerned, the process used for producing dimensions B or E, the
IT grade has to be reduced by one (a 20.27 percent change),
whereas for achieving the same goal, the IT grade for the process
used for the manufacture of dimension D or F has to be reduced
by two (a 60.00 percent change). In this case, both dimensions B
and E have similar effects and the CE team can choose either of
them for further analysis. If the selected option is the
manufacturing of dimension B, the CE team members will be
able to determine how, in practice, the IT grade could be reduced
by one and propose new process plans. They can then decide
which option will be the easiest to achieve (i.e., which option will
add the least cost/time to the products manufacturing cost). For
making this decision, a decision matrix, similar to the one
explained below, is used.
After consulting the process capacity database, the CE found
that the finish turning with a process capability of IT7-IT9 and
coarse grinding with a process capability of IT7-IT9 could be
used for the manufacture of dimensions B. On this basis, the
following options are generated:
#132
6
TABLE III
AN APPLICATION OF THE PROPOSED STRATEGY: INITIAL PROPOSAL
Target Value = 0.5 0.266
Retaining
Part Name
Shaft
Bearing
Bearing
Sleeve
Ring
Housing
Bearing
Sleeve
Bearing
Dimension
Sensitivity Factors
-1
-1
-1
-1
1.283
200
12.936
10
192.345
10
12.936
Basic Size
Purchased Part
Tolerance
Yes
Yes
Yes
0.019
0.032
0.032
N/A
Turning
N/A
Turning
Turning
Turning
N/A
IT7- IT11
N/A
IT7- IT11
IT7- IT11
IT7- IT11
N/A
N/A
N/A
IT9
N/A
IT9
IT9
IT9
N/A
N/A
0.056
N/A
0.019
0.056
0.019
N/A
Safety Factor
N/A
1.33
N/A
1.33
1.33
1.33
N/A
N/A
0.074
N/A
0.025
0.074
0.025
N/A
N/A
-20.27%
N/A
-60.00%
-20.27%
-60.00%
N/A
Required IT change
N/A
-1
N/A
-2
-1
-2
N/A
N/A
Yes
N/A
Yes
N/A
TABLE IV
AN APPLICATION OF THE PROPOSED STRATEGY:
DECISION MATRIX FOR COMPARING ALTERNATIVE OPTIONS.
Criteria
Weighting
0.43
1.00
0.03
1.00
0.54
1.00
Total
1.00
0.430
0.030
0.540
Option No 2
1.20
0.516
1.30
0.039
1.10
1.000
Option No 3
0.594
1.50
0.645
1.40
0.042
1.80
1.149
0.972
1.659
TABLE V
AN APPLICATION OF THE PROPOSED STRATEGY: FINAL REULS
Available +/- Tolerance = {0.266 - (0.019 + 0.032 + 0.032)} = 0.183
Proposed Manufacturing Process
N/A
Fine Turning
N/A
Turning
Turning
Turning
N/A
IT7- IT9
N/A
IT7- IT11
IT7- IT11
IT7- IT11
N/A
N/A
N/A
IT8
N/A
IT9
IT9
IT9
N/A
N/A
0.035
N/A
0.019
0.056
0.019
N/A
N/A
1.33
N/A
1.33
1.33
1.33
N/A
N/A
0.047
N/A
0.025
0.074
0.025
N/A
N/A
0.050
N/A
0.025
0.080
0.025
N/A
Preferred Tolerance
Effective Process Capability Index
N/A
N/A
N/A
Yes
1.33
Yes
1.43
Yes
1.33
N/A
N/A
Yes
1.43
Retaining
Shaft
Bearing
Bearing
Sleeve
Housing
Bearing
Sleeve
Bearing
N/A
Final Results
Part Name
Ring
Dimension
Sensitivity Factors
-1
-1
-1
-1
Purchased Part
Yes
Yes
Yes
Basic Size
1.283
200
12.936
10
192.345
10
12.936
Tolerance (+/-)
0.019
0.050
0.032
0.025
0.080
0.025
0.032
N/A
Yes
1.43
N/A
Yes
1.33
Yes
1.43
Yes
1.33
N/A
Preferred Tolerance
Effective Process Capability Index
N/A
N/A
N/A
#132
V. DISCUSSION
ACKNOWLEDGMENT
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
#132
[15] F. H. Speckhart, Mechanism design: Accounting for manufacturing
tolerances and costs in function generating problems, J. of Engineering
for Industry, 94, pp. 447-453. 1972.
[16] M. F. Spotts, Allocation of tolerances to minimize cost of assembly, J.
of Engineering for Industry, 95, pp. 762-764, 1973.
[17] D. Wide and E. Prentice, Minimum exponential cost allocation of surefit tolerances, J. of Engineering for Industry, 97, pp. 1395-1398, 1975.
[18] G. H. Sutherland and B. Roth, Calculation of tolerance based on a
minimum cost approach, J. of Engineering for Industry, 97, 283-286,
1975.
[19] P. F. Ostwald and J. Huang, A method for optimal tolerance selection,
J. of Engineering for Industry, 99, pp. 558-565, 1977.
[20] W. Michael and J. Siddall, The optimization problem with optimal
tolerance assignment and full acceptance, J. of Mechanical Design,
104, pp. 855-860, 1982.
[21] D. B. Parkinson, Assessment and optimization of dimensional
tolerances, Computer-Aided Design, 17, pp. 191-199, 1985.
[22] W. Lee, W and T. C. Woo, Optimum selection of discrete tolerances,
J. Mechanisms, Transmissions, and Automation in Design, 111, 243251, 1989.
[23] K. W. Chase and W. H. Greenwood, Least cost tolerance allocation for
mechanical assemblies with automated process selection,
Manufacturing Review, 3(1), pp. 49-59. 1990.
[24] W. H. ElMaraghy and H. A. ElMaraghy, Evaluation of cost-tolerance
algorithms for design tolerance analysis and synthesis, Manufacturing
Review, 1(3), pp. 168 -179, 1988.
[25] E. Balas, An additive algorithm for solving linear programs with zeroone variables, Operation Research, 13, pp. 517-546, 1965.
[26] G. Taguchi, E. A. Elsayed and T. C. Hsiang, Quality engineering in
production systems, McGraw-Hill book Co., 1989.
[27] K. Krishnaswami and R. W. Mayne, Optimizing tolerance allocation
based on manufacturing cost and quality loss, Advances in Design
Automation, 11, pp. 211-217, 1994.
[28] R. Sderberg, Tolerance allocation considering customer and
manufacturer objectives, Advances in Design Automation 1993. 2,
pp. 149-157, 1993.
[29] R. Sderberg, Robust design by tolerance allocation considering
quality and manufacturing cost, Advances in Design Automation 1994, 1, pp. 219-226, 1994.
[30] H. R. Choi and M. H. Park, Optimal tolerance allocation with loss
function, J. Manufacturing Science and Engineering, 122(3), pp. 529535. 2000.
[31] K. C. Kapur, An approach for development of specifications for
quality improvement, Quality Engineering, 1(1), pp. 63-77, 1989.
[32] Y. Fathi, R. O. Mittal, J. E. Cline and M. Martin, Alternative
manufacturing sequences and tolerance buildup: A point of view and a
case study Int. J. Prod. Res., 35(1), pp. 123-136, 1997.
8
[33] A. Jeang, An approach of tolerance design for quality improvement
and cost reduction, Int. J. Prod. Res., 35(5), pp. 1193-1211, 1997.
[34] D. Janakiram, L. V. Prasad and U. R. K. Rao, Tolerancing of parts
using an expert system, Int. J. of Advanced Manufacturing Technology,
4, pp. 157-167, 1989.
[35] E. Dupinet, M. Balazinski and E. Caogala, Tolerance allocation based
on fuzzy logic and simulated annealing, J. of Intelligent
Manufacturing, 7, pp. 487-497, 1996.
[36] P. Kopardekar and S. Anand, Tolerance allocation using neural
network, Int. J. of Adv. Manuf. Technol., 10, pp. 269-276, 2005.
[37] M. Iannuzzi and E. Snadgren, Tolerance optimization using genetic
algorithms: Benchmarking with manual analysis, Paper presented at
the 4th CIRP Seminar on Computer Aided Tolerancing, Tokyo. April 56, 1995.
[38] G. Bai, C. Zhang and B. Wang, Optimization of marching datum
selection and machining tolerance allocation with genetic algorithms, I.
J. of Prod. Research, 38, pp. 1407-1424, 2000.
[39] M. H. Gadallah and H. A. ElMaraghy, The tolerance optimization
problem using a system of experimental design, Advances in Design
Automation, 1(4), 251-264, 1994.
[40] R. G. Wilhelm and S. C. Y.Lu, Tolerance synthesis to support
concurrent engineering, Annals of the CIRP, 41(1), pp. 197-200, 1992.
[41] L. E. Farmer, Function Oriented Dimensioning Enhances Concurrent
Engineering Performance, Proceedings ACME93, IE Aust., 1993.
[42] M. N. Islam, Concurrent engineering offers the best solution to
tolerance allocation problems, New Engineer J., 6 (1), pp. 14-15, 2003.
[43] M. Imai, Kaizen: The key to Japans competitive success, McGraw-Hill
Publ. Co. 1986.
[44] J. R. Dixon and C. Poli, Engineering design and design for
manufacturing - A structured approach, Massachusetts: Field Stone
Publ., 1995.
[45] H. G. Conway, Engineering tolerances. 3rd Ed. London: Sir Isaac
Pitman and Sons. 1966.
[46] C. A. Gladman, Geometric analysis of engineering designs. 2nd Ed.,
Sydney: Australian Trade Publ. Pty. Ltd., 1972.
[47] D. H. Besterfield, Quality control, 2nd Ed. Prentice-Hall, 1986.
[48] T. Pyzdek, What every engineer should know about quality control,
New York: Marcel Dekker. 1989.
[49] P. F. Ostwald, Engineering cost estimating. 3rd Ed., Prentice Hall, 1992.
[50] Tool and manufacturing engineers handbook, Vol. 6: Design for
Manufacturability, SME, 1993.
[51] Tool and Manufacturing Engineers Handbook, Vol. 1: Machining,
SME, 1983.