Documente Academic
Documente Profesional
Documente Cultură
DOI 10.1007/s10845-007-0030-z
Received: May 2005 / Accepted: August 2006 / Published online: July 2007
Springer Science+Business Media, LLC 2007
Abstract In this paper, we consider the problem of buffer Introduction and review of the literature
space allocation for a tandem production line with unreli-
able machines. This problem has various formulations all Buffer allocation problems arise in a wide range of appli-
aiming to answer the question: how much buffer storage to cations: automatic transfer lines, production and assembly
allocate between the processing stations? Many authors use systems or communication networks. These exist in various
the knapsack-type formulation of this problem. We inves- forms, all aiming to answer the question: how much buffer
tigate the problem with a broader statement. The criterion storage to allocate between the processing stations? This
depends on the average steady-state production rate of the question is important because the buffers may prevent block-
line and the buffer equipment acquisition cost. We evaluate ing and/or starving of stations, thus drastically influencing
black-box complexity of this problem and propose a hybrid the efficiency of the whole system. This is especially impor-
optimization algorithm (HBBA), combining the genetic and tant for JIT environment which has as objective to reduce
branch-and-bound approaches. HBBA is excellent in com- the inventory to as close to zero as possible. Therefore, we
putational time. HBBA uses a Markov model aggregation need to limit radically the stock by bounding the storage
technique for goal function evaluation. Nevertheless, HBBA space between stations. In the context of Kanban policy, the
is more general and can be used with other production rate buffer allocation problem is equivalent to the calculation of
evaluation techniques. the minimal number of Kanbans.
An excellent illustration of the value to industry in
Keywords Production line Buffer allocation NP solving problems of this type is given by Patchong,
completeness Black-box complexity Genetic algorithm Lemoine, and Kern (2003). The authors demonstrate how
Branch-and-Bound method Hybrid algorithm methods for buffer allocation in designing PSA Peugeot
Citron car-body shop yielded substantial profits. The
practical importance of the optimization tools for buffers
allocation was also demonstrated by Tempelmeier (2003).
A detailed analysis of mathematical models describing the
A. Dolgui (B)
effect of the buffer storage may be found in the follow-
Division for Industrial Engineering and Computer Sciences, Ecole
des Mines de Saint Etienne, 158, Cours Fauriel, ing books (Altiok, 1996; Buzacott & Shanthikumar, 1993;
42023 Saint-Etienne Cedex 2, France Gershwin, 1994) and comprehensive surveys (Buzacott &
e-mail: dolgui@emse.fr Hanifin, 1978; Dallery & Gershwin, 1992; Papadopoulos &
Heavey, 1996).
A. V. Eremeev
Discrete Optimization Laboratory, Omsk Branch of Sobolev In this paper, we consider a tandem production line
Institute of Mathematics, Omsk, Russia (see Fig. 1), where the parts are moved from one machine to
e-mail: eremeev@ofim.oscsbras.ru the next by some kind of transfer mechanism. The machines
are subject to breakdowns: when a breakdown occurs, the cor-
V. S. Sigaev
Department of Mathematics, Omsk State University, Omsk, Russia responding machine is unusable for a random repair period.
e-mail: SigVS@yandex.ru The machines are separated by finite buffers. The parts are
123
412 J Intell Manuf (2007) 18:411420
123
J Intell Manuf (2007) 18:411420 413
Note: The GA for finding the initial approximate solution instance but may collect information about the unknown
was chosen after testing various other versions of GAs and parameters only through goal function evaluations.
several tabu search (TS) algorithms we developed. One of the Let us call the number of tentative solutions examined by a
most efficient versions of the TS algorithm was one which search algorithm (randomized or deterministic), until the opti-
used a random neighborhood space and constant length of mum is found, the optimization time. The complexity of a
tabu-listsee e.g. (Glover and Laguna, 1997). However, the black-box optimization algorithm is defined as the expected
selected GA achieved better results. The superiority of the (average) optimization time for the worst-case instance, which
GA versus TS is partly explained by the effects of a pop- is a function of the problem size. This approach is well suited
ulation that allows the GA to adaptively search in different for analysis of the problem hardness for a wide class of mod-
areas of the decision space. This feature proved to be helpful ern heuristic methods, such as genetic algorithms, simulated
in the Nested Partitions algorithm of Shi and Men (2003), annealing, evolutionary strategies, tabu search, etc. (see e.g.
which is also based on information accumulation helping to Reeves, 1993), where the search process is mainly directed by
concentrate the search in the most promising areas. the goal function values of already visited solutions.
Taking into account that the algorithm HBBA is coupled Proposition 1 For the buffer space allocation problem for a
with an approximate production rate evaluation algorithm, it line consisting of two machines separated by a finite buffer,
becomes an approximation algorithm as well. Nevertheless, the expected optimization time of any black-box optimization
the precision of the HBBA is provably close to that of goal algorithm is lower bounded by d1 /2 + 1.
function evaluation.
Proof Let the integer upper bound d1 for the buffer size h 1
be given. In order to construct a hard case for optimization,
Optimization problem properties one can define such functions R(V ) and J (h 1 ) for this line
that (h 1 ) will be constant (let it be 0) for all integer values
Criterion of h 1 , except at one point, where (h 1 ) takes its maximal
value (assume that here (h 1 ) equals 1). This is based on the
Let us introduce the following additional notation: fact that function V (h 1 ) is strictly increasing on [0, d1 ] (this
Tam amortization time of the line (line life cycle); can be shown using the closed-form expressions e.g. from
Coillard & Proth, 1984), so the maximum of (h 1 ) may be
R(V ) revenue for the sold production per time unit;
hidden in any point 0, 1, . . ., d1 .
J (H ) buffers acquisition cost for configuration H ; Now we can apply the Yaos minimax principle (see e.g.
di maximal admissible size for buffer i, i = Motwani & Raghavan, 1995), which gives the lower bounds
1, 2, . . ., N . on complexity of the randomized black-box optimization
In this paper, we deal with the following criterion: algorithms through the analysis of deterministic algorithms
for the same problem. Let us limit the set of problem inputs
Max (H ) = Tam R(V (H )) J (H ). (1) to those where (h 1 ) takes only values 0 and 1 for h 1 in
The functions R(V ) and J (H ) are assumed monotone and {0, 1, . . ., d1 } (this will further simplify the problem). In our
non-decreasing. These functions may incorporate some situation, for any fixed value of d1 , the Yaos minimax prin-
penalties, fixed costs for different standard buffer sizes, over- ciple implies that the optimization time of any black-box
production price reduction, etc. Function (H ) is to be maxi- algorithm for its worst-case function (h 1 ) is lower-bounded
mized subject to the constraints h 1 d1 , h 2 d2 , . . ., h N by the expected time of the worst-case optimal deterministic
dN . black-box algorithm (for the same value of d1 ) working on
the inputs with any given probability distribution of function
Problem complexity (h 1 ).
Taking (h 1 ) uniformly distributed on the set of functions
In Appendix, we give a proof of NP-hardness for a simple with a single maximum (equal to 1) in {0, 1, . . ., d1 }, we con-
case of parallel-serial lines. The problem considered in the clude that the expected optimization time of any worst-case
main body of this paper is more difficult to analyze, since optimal deterministic black-box algorithm is d1 /2 + 1. Thus,
it contains two arbitrary non-decreasing functions R(V ) and by the Yaos minimax principle, the expected optimization
J (H ), which makes the usual complexity analysis (as e.g. in time of any black-box algorithm for our problem is lower
Garey & Johnson, 1979) not quite adequate. bounded by d1 /2 + 1 (the obtained lower bound is tight, as
Therefore, we analyzed the complexity of the problem it follows from Proposition 2 of Droste et al., 2006).
in terms of black-box optimization (see e.g. Droste, Jansen, Finally, our Proposition 1 demonstrates that the black-box
& Wegener, 2006). Black-box optimization is used when we complexity of buffer space allocation problem can be arbi-
do not have an access to the specific parameters of the given trarily large as we increase the maximal admissible buffer
123
414 J Intell Manuf (2007) 18:411420
123
J Intell Manuf (2007) 18:411420 415
GA using LOBBA . The general scheme of the GA coin- mization procedure of GA and for the full-scale problem after
cides with that of the genetic algorithm proposed in (Dolgui running the GA.
et al., 2002), except for the local optimization procedure. A
solution (individual) is presented by a vector of components Some remarks on precision of BBA
(genes), where each gene gives the size of the corresponding
buffer. The solutions of the initial population are randomly To implement the BBA with such bounds we need to compute
generated according to an a priori defined probability dis- the functions (S(F, a, b, g, j)) and (H ) exactly, which
tribution. At each iteration, a couple of new solutions are is problematic, since no exact method is known to evaluate
obtained, and they replace a couple of unpromising solu- the production rate V (H ). Instead, let us introduce the func-
tions chosen in the current population (thus the population tions (S(F, a, b, g, j)) and (H ) defined by analogy with
size remains constant). The best solution is returned when (S(F, a, b, g, j)) and (H ), respectively, except for the
the GA stops. production rate V which is now substituted by its approxi-
Construction of a new pair of solutions (offspring) starts mate value, computed via a Markov-model aggregation heu-
with choosing a pair of parents from the current population ristic.
by the means of a probabilistic selection operator. We use the We can formulate an a priori precision of the BBA in terms
s-tournament selection operator which randomly chooses s of deviations of (S(F, a, b, g, j)) from (S(F, a, b, g, j))
individuals from the current population and selects the best and (H ) from (H ). Let us consider a general case and
one as a parent. Then the standard one-point crossover opera- assume that for the problem being considered, the values
tor (see e.g. Reeves, 1993) replaces some coordinates (genes) and are such that
of one parent with the values taken from the corresponding
|(H ) (H )| for all H D,
positions of the other. This is done with a fixed crossover
probability Pcross (otherwise the crossover has no effect). and for any 5-tuple (F, a, b, g, j) holds
After crossover, the strings undergo mutation, where the
(S(F, a, b, g, j)) (S(F, a, b, g, j)) .
genes are randomly altered: the size of each buffer receives
a uniformly distributed random variation of not more than The following simple proposition bounds the precision of the
units (parameter is chosen experimentally before the run BBA.
of the GA).
Proposition 2 If H is an optimal solution and H is the
The local optimization heuristic LOBBA is applied to the
output of the BBA, then
obtained solutions before they are added into population.
This modification of the GA using the LOBBA has turned (H ) (H ) + max(, ).
out to be more advantageous than those suggested in (Dolgui Proof Let us assume H = H . This means that one of the
et al., 2002). pruned nodes of the branching tree was associated with a set
of solutions S(F, a, b, g, j), containing the solution H . If
The hybrid algorithm HBBA this node is a leaf, then (H ) (H ) 0, otherwise
(H ) (S(F, a, b, g, j)) 0.
The hybrid algorithm HBBA consists in starting the GA
In the first case we have
before the BBA for finding an approximate solution (see
Fig. 2). This solution is used further in the BBA, defining (H ) + ((H ) ) > 0,
an initial lower bound (L B). If no improving solutions are
(H ) (H ) 2,
found in the BBA, then this solution is returned as a result of
the HBBA: and in the second case
(H )+((H ) ) > (H )
(S( f, a, b, g, j)) 0,
Algorithm HBBA
therefore,
1. Obtain solution H using GA with LOBBA . (H ) (H ) + .
2. Start BBA N from node (, 0, d1 , , 1) with L B = (H ).
3. If BBA N finds H , (H ) < (H ) then return H ,
In our case = , so (H ) (H ) 2. Note: the
otherwise return H .
complete enumeration of all feasible solutions in the worst
case also yields an error 2. Thus, the BBA achieves the best
As mentioned before, the BBA is used in HBBA both for possible precision in some sense (it does not introduce any
solving the one-dimensional sub-problems in the local opti- additional errors).
123
416 J Intell Manuf (2007) 18:411420
123
J Intell Manuf (2007) 18:411420 417
2,5
We were unable to compare the proposed algorithms with
the CEM on the whole set of test series due to the immense 2
computational time required for the CEM for the larger prob-
lems. However, it turned out that for the series from G5_20 to 1,5
123
418 J Intell Manuf (2007) 18:411420
Table 3 Running times of the Pure BBA and the solutions obtained In Tables 4 and 5, we list the timetable for the movements
Problem H BBA i of parts. The part arrival times are given in column A and
part departure times are given in column D.
vp6_3 (0,2,5,3) .3359 .000 The tables show that the number of parts processed by the
vp6_4 (0,0,3,1) .3602 .010 line till time 6 in the first case (4 parts) is greater than in the
vp6_5 (0,1,3,1) .3782 .010 second one (3 parts), so the monotonicity property does not
vp6_6 (0,1,3,2) .3946 .020 hold.
vp6_7 (0,1,4,2) .4094 .020 Of course this example does not imply that the monoto-
vp6_8 (0,2,4,2) .4215 .030 nicity of V (H ) can not be established by some other method,
vp6_9 (0,2,5,2) .4334 .040 different from the sample path. Nevertheless, using the sim-
vp6_10 (5,3,2,0) .4427 .050
ulation (see Dolgui, 1993 for details of the simulation algo-
rithm) on the line with the same structure but with other
vp7_3 (2,1,0,0) .4515 .010
parameters we found that V (H ) increases when h 2 is reduced
vp7_4 (2,1,1,0) .4605 .020
(with statistical significance level .05). The parameters that
vp7_5 (3,1,1,0) .3359 .030
we used in this experiment were i = i = 1 for all i =
vp7_6 (3,2,1,0) .3602 .050
1, 2, 3, 4, u 1 = u 2 = 1, u 3 = .5, u 4 = 2, h 1 = 10, 000,
vp7_7 (4,2,1,0) .3782 .071
h 2 = 50, h 3 equals 10, 11 and 20, h 4 = .
vp7_8 (4,3,1,0) .3946 .111
vp7_9 (5,3,1,0) .4094 .130
vp7_10 (5,3,2,0) .4215 .170
Conclusions
performance analysis) discussed above is also applicable to A buffer allocation problem for unreliable tandem lines was
these lines (Dolgui et al., 2002). considered. We suggested a method named HBBA. This
However, the proposed BBA is harder to generalize, method has a precision comparable to that of the complete
because the series-parallel lines are significantly harder with enumeration. HBBA is based on the branch-and-bound
respect to finding appropriate bounds. The UB used in this approach, complemented by a genetic algorithm for finding
paper is based on monotonicity of the function V (H ) which initial solutions. For the evaluation of production rate of ten-
was proved by the sample path approach. But, for the series- tative design decisions we used a Markov-model aggregation
parallel lines, we found a counterexample with 4 machines method, however, our optimization technique is quite general
(see Fig. 5), which shows that the sample path approach is not and it may be used for other evaluation methods and models,
applicable for the proof of monotonicity of function V (H ). provided that V (H ) is a non-decreasing function of H .
In this counterexample, we use the following assumptions: The computational experiments showed that the usage
of an initial solution, obtained by the genetic algorithm, in
the branch-and-bound procedure, can shorten the total run-
A part goes from Buffer 1 to Machine 2, if Machine 2 is
ning time as compared to the pure branch-and-bound algo-
free, otherwise it goes to Machine 4,
rithm. Another new feature of the proposed hybrid algorithm
all machines never fail,
HBBAthe insertion of the one-dimensional Branch and
u 1 = u 2 = u 4 = 1, u 3 = 1/2,
Bound algorithm into the genetic algorithm for local optimi-
the last buffer is inexhaustible.
zationturned out to be advantageous.
Further research might address the production lines with
Let us compare two following examples. Assume that the series-parallel structure. An efficient upper bound for this
sizes of the first and second buffers in the first case are zero: type of line needs to be found. Moreover, it would be interest-
h 1 = h 2 = 0, and in the second case: h 1 = 0, h 2 = 1. ing to use the statistical information collected in the genetic
Buffer 1 Buffer 3
Machine 1
Machine 4
123
J Intell Manuf (2007) 18:411420 419
Table 4 First case timetable Proposition 3 The problem of finding the buffer space allo-
n1
Machine 1 Machine 2 Machine 3 Machine 4 Buffer 3 cation vector H = (h 1 , h 2 , . . ., h n1 ) Z + minimizing
n1
A D A D A D A D A the criterion b j h j , subject to constraints V (H ) V 0 ,
j=1
0 1 1 2 2 4 4 h 1 d1 , h 2 d2 , . . ., h n1 dn1 for line with simple
1 2 2 4 4 6 6 structure and rational weights b1 , . . ., bn1 , V 0 and i , i ,
2 3 3 4 4 u i , i = 1, . . ., m is NP-hard.
3 4 4 6 6 8 8 Proof Consider a special case of the problem with m =
4 5 5 6 6 2(n 1), where each chain consists of two sequential
machines indexed i, i + 1, i = 1 mod 2. Let j = (i + 1)/2
be the index of the buffer between machines i, i + 1. Assume
Table 5 Second case timetable also that i = 2u i , i = 4u i for all i = 1, . . ., m, d1 =
Machine 1 Machine 2 Buffer 2 Machine 3 Machine 4 Buffer 3 d2 = = dn1 = 1 and u i = u i+1 for all i = 1
mod 2. By means of formulas from (Dolgui, 1993) in case
AD AD AD AD AD A
h j = 1 we obtain: V j = 8u 2 j /13, where V j is the through-
0 1 1 2 2 2 2 4 4 put of chain with buffer j; and in case h j = 0 we have:
1 2 2 3 3 4 4 6 6 V j = u 2 j /2. The throughput of the whole system is V (H ) =
2 3 3 4 4 6 6 8 8 V1 + V2 + + Vn1
and thus all necessary system parame-
3 4 4 6 6 8 8 10 10 ters are computable in polynomial time in length of problem
4 5 5 6 6
input encoding.
To prove the NP-hardness of the described problem, we
reduce the partition problem (Garey & Johnson, 1979) to
the described problem. The optimization version of partition
algorithm for finding the order of branching that suits a given problem is NP-hard and it can be formulated as follows:
problem.
N
Min aj yj,
Acknowledgements The research is supported by INTAS (projects
j=1
00-217 and 03-51-5501). The authors thank Chris Yukna for checking
the English of this paper.
N
1
N
aj yj aj,
2
j=1 j=1
any stronger results in this direction for the case of tandem V (H ) = V j = ajh j + aj = ajh j + V0
3
j=1 j=1 j=1 j=1
production lines considered in this paper; however, the knap-
sack-type problem for series-parallel lines turned out to be 1
n1
mial in length of problem input data. It will be sufficient for us The objective functions are identical.
to present such a procedure only for the production systems
with simple structure. By system with a simple structure, we
mean a system consisting of parallel chains only, where each References
chain has at most two sequential machines, there are no links
between the chains, and all chains start at Machine 1 and Altiok, T. (1996). Performance analysis of manufacturing systems. New
terminate at the last machine. York: Springer.
123
420 J Intell Manuf (2007) 18:411420
Buzacott, J. A., & Hanifin, L. E. (1978). Models of automatic transfer Glover, F., & Laguna, M. (1997). Tabu search. Kluwer Academic
lines with inventory banks: A review and comparison. AIIE Trans- Publishers.
actions, 10(2), 197207. Helber, S. (2001). Cash-flow-oriented buffer allocation in stochastic
Buzacott, J. A., & Shanthikumar, J. G. (1993). Stochastic models of flow lines. International Journal of Production Research, 39, 3061
manufacturing systems. New Jersey: Prentice Hall. 3083.
Coillard, P., & Proth, J. M. (1984). Effet des stocks tampons dans une Inman, R. R. (1999). Empirical evaluation of exponential and indepen-
fabrication en ligne. Revue belge de Statistique, dInformatique et dence assumptions in queueing model of manufacturing systems.
de Recherche Oprationnelle, 24(2), 327. Production and Operations Management, 8(4), 409432.
Dallery, Y., David, R., & Xie, X. (1989). Approximate analysis of trans- Jafari, M. A., & Shanthikumar, J. G. (1989). Determination of opti-
fer lines with unreliable machines and finite buffers. IEEE Transac- mal buffer storage capacities and optimal allocation in multistage
tions on Automatic Control, 34, 943953. automatic transfer lines. IIE Transactions, 21(2), 130135.
Dallery, Y., & Gershwin, S. B. (1992). Manufacturing flow line sys- Levin, A. A., & Pasjko, N. I. (1969). Calculating the output of transfer
tems: A review of models and analytical results. Queuing systems lines. Stanki i Instrument, 8, 810 (in Russian).
theory and applications, 12, 394 (Special Issue on Queuing Model Li, J. (2005). Overlapping decomposition: A system-theoretic method
of Manufacturing Systems). for modeling and analysis of complex manufacturing systems. IEEE
De Koster, M. B. M. (1987). Estimation of line efficiency by Transactions on Automation Science and Engineering, 2(1), 4053.
aggregation. International Journal of Production Research, 25, Motwani, R., & Raghavan, P. (1995). Randomized algorithms. Cam-
615626. bridge University Press.
Dolgui, A. (1993). Analyse de performances dun atelier de produc- Papadopoulos, H. T., & Heavey, C. (1996). Queueing theory in man-
tion discontinue: mthode et logiciel, Research Report INRIA 1949, ufacturing systems analysis and design: A classification of models
44 pp. for production and transfer lines. European Journal of Operational
Dolgui, A., Eremeev, A., Kolokolov, A., & Sigaev, V. (2002). A genetic Research, 92, 127.
algorithm for the allocation of buffer storage capacities in a produc- Patchong, A., Lemoine, T., & Kern, G. (2003). Improving car body
tion line with unreliable machines. Journal of Mathematical Model- production at PSA Peugeot Citroen. Interfaces, 33(1), 3649.
ling and Algorithms, 1(2), 89104. Reeves, C. R. (1993). Modern heuristic techniques for combinatorial
Dolgui, A. B., & Svirin, Y. P. (1995). Models of evaluation of prob- problems. New York, NY: John Wiley & Sons.
abilistic productivity of automated technological complexes. Vesti Shi, L., & Men, S. (2003). Optimal buffer allocation in production lines.
Akademii Navuk Belarusi: phisika-technichnie navuki, 1, 5967 (in IIE Transactions, 35, 110.
Russian). Srensen, K., & Janssens, G. K. (2004). A Petri net model of a contin-
Droste, S., Jansen, T., & Wegener, I. (2006). Upper and lower bounds uous flow transfer line with unreliable machines. European Journal
for randomized search heuristics in black-box optimization. Theory of Operational Research, 152, 248262.
of Computing Systems, 39(4), 525544. Spinellis, D., & Papadopoulos, C. (2000). Stochastic algorithms for
Dubois, D., & Forestier, J. P. (1982). Productive et en-cours moyens buffer allocation in reliable production lines. Mathematical Prob-
dun ensemble de deux machines spares par une zone de stockage. lems in Engineering, 5, 441458.
RAIRO Automatique, 16(2), 105132. Tempelmeier, H. (2003). Practical considerations in the optimization
Garey, M. R., & Johnson, D. S. (1979). Computers and intractabil- of flow production systems. International Journal of Production
ity. A guide to the theory of NP-completeness. San Francisco: W.H. Research, 41(1), 149170.
Freeman and Company. Terracol, C., & David, R. (1987). An aggregation method for perfor-
Gershwin, S. B. (1987). An efficient decomposition method for the mance valuation of transfer lines with unreliable machines and finite
approximate evaluation of tandem queues with finite storage space buffers. Proceedings of the IEEE International Conference on Robot-
and blocking. Operations Research, 35(2), 291305. ics and Automation, 13331338.
Gershwin, S. B. (1994). Manufacturing systems engineering. New Jer- Vouros, G. A., & Papadopoulos, H. T. (1998). Buffer allocation in unre-
sey: Prentice Hall. liable production lines using a knowledge based system. Computers
Gershwin, S. B., & Schor, J. E. (2000). Efficient algorithms for buffer and Operations Research, 25(12), 10551067.
space allocation. Annals of Operations Research, 93, 117144. Yamashita, H., & Altiok, T. (1998). Buffer capacity allocation for a
Glasserman, P., & Yao, D. D. (1996). Structured buffer-allocation prob- desired throughput in production lines. IIE Transactions, 30, 883
lems. Journal of Discrete Event Dynamic Systems, 6, 942. 891.
123
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.