Você está na página 1de 12

European Journal of Operational Research 114 (1999) 557568

Theory and Methodology

A sequential heuristic procedure for one-dimensional cutting


Miro Gradisar
a

a,b,*

, Miroljub Kljajic a, Gortan Resinovic b, Joze Jesenko

Faculty of Organizational Sciences, University of Maribor, 4000 Kranj, Kidri


ceva ulica 55 A, Slovenia
b
Faculty of Economics, University of Ljubljana, 1000 Ljubljana, Kardeljeva ploscad 17, Slovenia
Received 10 September 1997; accepted 23 March 1998

Abstract
The article examines the Sequential Heuristic Procedure (SHP) for optimising one-dimensional stock cutting when
all stock lengths are dierent. In order to solve a bicriterial multidimensional knapsack problem with side constraints a
lexicographic approach is applied. An item-oriented solution was found through a combination of approximations and
heuristics that minimize the inuence of ending conditions leading to almost optimal solutions. The computer program
CUT was developed, based on the proposed algorithm. Two sample problems are presented and solved. A statistical
analysis of parameters that aect material utilisation was also made. 1999 Elsevier Science B.V. All rights reserved.
Keywords: Cutting; Heuristics; Optimisation

1. Introduction
The problem of one-dimensional stock cutting
occurs in many industrial processes [13] and
during the past few years it has attracted an increasing attention of researchers from all over the
world [4]. Yet attention has been mostly focused
on the solution to the problem in cases with the
stock of the same length or with a few dierent
standard lengths. However, when the stock lengths
are all dierent, generally acceptable solutions to
this problem have not appeared in the literature so
far. The purpose of this article is to propose a

Corresponding author. E-mail: miro.gradisar@fov.unimb.si.

solution that will generalise and improve our earlier solution [5] where a practical problem in the
clothing industry is considered.
Most standard problems related to one-dimensional stock cutting are known to be NPcomplete. However, in many cases the problems
can be modelled by means of mathematical programming and a solution can be found by using
approximate methods and heuristics. Our objective is to design a plan of one-dimensional cutting
of a certain number of pieces of dierent lengths
(stock lengths), into a large number of short pieces
(order lengths), which will minimize the overall
trim loss considering dierent conditions that may
appear in practice.
Using Dycko's typology [6] our integer cutting
stock problem can be described, in the cases with

0377-2217/99/$ see front matter 1999 Elsevier Science B.V. All rights reserved.
PII: S 0 3 7 7 - 2 2 1 7 ( 9 8 ) 0 0 1 4 0 - 4

558

M. Gradisar et al. / European Journal of Operational Research 114 (1999) 557568

enough required material available, as 1/V/D/M,


where 1 refers to one-dimensional problem, V
means that all items must be produced from a selection of large objects, D means that all large
objects are dierent and M indicates many small
items of many dierent dimensions. In other cases
we can describe it as 1/B/D/M, where B means all
large objects and a selection of small items.
Dycko classies the solution of integer cutting
stock problems into two groups: item-oriented and
pattern-oriented approach. Item-oriented approach
is characterised by individual treatment of every
item to be cut. In the pattern-oriented approach, at
rst, order lengths are combined into cutting patterns, for which in a succeeding step the frequencies are determined that are necessary to
satisfy the demands. The literature abounds in
pattern-oriented solutions based on a hybrid algorithm, which was developed by Gilmore and
Gomory [7,8]. However, a pattern-oriented approach is possible only when the stock is of the
same length or of the several standard lengths. Our
solution is an item-oriented one because all stock
lengths are dierent and frequencies cannot be
determined.
Item-oriented solution can be based on exact
methods or on approximation algorithms [6]. For
the 1/V/D/M or 1/B/D/M type of problem there is
no exact method which could nd an optimal itemoriented solution within reasonable time limits.
The exact algorithm that would provide a solution
without the unnecessary trim loss would be useless
due to high time complexity. Therefore, a solution
in the form of approximation algorithm has to be
found.
In the literature the authors have not come
across an approximation algorithm that would be
directly applicable to the described problem.
Similar problems are solved by Sequential Heuristic Procedure (SHP). A similar kind of problem
is the classical ``bin packing problem'', which can
be solved for example by the ``First Fit Decreasing'' (FFD) or ``Best Fit Decreasing'' (BFD)
SHP [9]. However, the basic feature of this type
of problem is that all stock lengths are the same.
Gilmore and Gomory's knapsack method for
pattern generation can also not be used directly.
Their pattern-oriented method is based on as-

sumption that there are only few standard stock


lengths and the optimal combinations for each of
them are independent. We have to adapt their
method to the case when all stock lengths are
dierent. An input data for the next step are dependent on results of the previous step and the
optimal combinations cannot be calculated simultaneously. An extensive review of the relevant
literature up to 1992 is given in [10].
The primary advantage of SHP is its ability to
control factors other than trim loss and to eliminate
rounding problems [11,12] by working only with
integer values. The major disadvantage of an SHP
is that it may generate a solution that has a greatly
increased trim loss because of the so-called ending
conditions. We have developed such an SHP, which
minimizes the inuence of ending conditions.
The objective of the optimisation is material
cost reduction with the plan of stock cutting optimisation not being too complex and the cost of
cutting as low as possible. So the solution oers a
compromise between the plan complexity and trim
loss by setting parameter Y. Y determines the
number of dierent order lengths that will be cut
out of one stock length. The statistical analysis
given in Section 4 reveals that the value of Y
should be between 1 and 4. For Y 4, the trim loss
becomes so low, that any further reduction would
probably not outweigh the extra eorts of cutting.
The article describes the cutting problem and
solution development in the form of a computer
program. Two examples of a practical implementation of the program are given. At the end the
inuence of parameters aecting utilisation of
stock is analysed.
2. Problem denition and a formal model
For every customer order a certain number of
stock lengths is available. In general all stock
lengths are dierent. We consider the lengths as
integers. If they are not originally integers we assume that it is always possible to multiply them
with a factor and transform them to integers. It is
necessary to cut a certain number of order lengths
into required number of pieces. The following
notation is used:

M. Gradisar et al. / European Journal of Operational Research 114 (1999) 557568

order lengths; i 1,. . .,n,


required number of pieces of order length si ,
stock lengths; j 1,. . .,m,
number of pieces of order length si having
been cut from stock length j.

si
bi
dj
xij

Two cases are possible:


Case 1: the order can be fullled as the abundance of material is in stock.
(1) min
(2) s.t.
(3)
(4)
(5)
(6)

Pm

(minimize trim loss which is


j1 tj
smaller than UB (upper bound for the
trim loss))
P
n
8j (knapsack
i1 si  xij dj dj
constraints),
Pm
8i (demand conj1 xij bi
straints the numbers of pieces are all
xed),
Pm
(maximum number of
j1 uj 6 1
residual lengths which are larger than
max
Pn si ),
8j (maximum
i1 yij 6 Y 6 n
number of dierent order lengths for
a stock length),
UB 6 max si , xij P 0, integer "i, j,
tj P 0, "j, dj P 0, "j.

For the above model the following functions


are used:

zj

0 if xij 0 8i;
1 otherwise;

to indicate whether stock length j is used in the


cutting plan;

yij

0 if xij 0;
1 otherwise;

to indicate whether order length i is cut from roll j;



uj

1 if zj 1 ^ dj > max si ;
0 otherwise;

to indicate whether the trim loss relating to stock


length j is greater than the longest order length;


tj

dj
0

559

if zj 1 ^ dj 6 UB;
otherwise:

tj indicates the extent of the trim loss relating to


stock length j.
Case 2: the order cannot be fullled entirely due
to shortage of material in stock.
In case 2 denition of the problem depends on
the way in which the uncut pieces are distributed
by order lengths. The possibilities that cover most
situations in practice are explained as follows:
2.1. Distribution of uncut pieces by order
lengths is not important.
(1)

min

(2)
(3)

s.t.

(4)
(5)

Pn

(minimize sum of uncut


i1 Di  si
order lengths)
same
Pm as in case 1,
8i (demand
j1 xij bi Di
constraints),
same as (5) in case 1,
Di P 0, integer "i.

2.2. Distribution of uncut pieces by order


lengths is important. This means that some order
lengths are more important and therefore we try to
full the order for those lengths rst. Besides trim
loss also distribution of uncut order lengths has to
be considered. This leads to a bicriterial problem.
As both criteria, the trim loss and distribution
could be in conict, a decision on priority is to be
made. There are two possibilities:
2.2.1. Trim loss is more important and distribution is approximate. We assume that:
(a) uncut pieces belong to all order lengths and
are evenly distributed,
(b) uncut pieces belong to the shortest order
lengths,
(c) uncut pieces belong to the longest order
lengths.
(1)

min

(2)

min

(3)
(4)

s.t.

Pn

(minimize number of
i1 Di
uncut order lengths and satisfy
condition (a) or (b) or (c)),
P
m
(minimize trim loss
j1 dj
which is smaller than max si ),
same as (2) in case 1,
same as (3) in case 2.1,

560

(5)
(6)

M. Gradisar et al. / European Journal of Operational Research 114 (1999) 557568

same as in case 1,
same as (5) in case 2.1.

2.2.2. Distribution is more important, so the


choice of any a priori distribution is assumed. In
this case the number of required pieces of a particular order length is being reduced as long as
enough material is available. After this a new
cutting plan is made. The problem becomes onecriterial again and the minimum trim loss is determined, given a precise a priori distribution, with
the following condition:
Pn
(minimize number of uncut
(1) min
i1 Di
order lengths at a determined a priori
distribution)
(2) s.t.
same as in case 1,
(3)
same as in case 2.1,
(4)
same as (5) in case 1,
(5)
same as in case 2.1.
Condition (6) determines exact distribution of
uncut pieces. Let us consider dierent possibilities
of the condition (6) for (a), (b), and (c) of 2.2.1:
(a) 0 6 Di Dk 6 1; k 1; . . . ; n 8i; k:
Set si is divided into two subsets si1 and si2 . Subset si1 consists of the lengths for which: Di > 0,
while subset si2 consists of the lengths for which:
Di 0. The condition (6) for (b) and (c) is dened as:
(b) max si1 6 min si2 ,
(c) min si1 P max si2 .
Unutilised stock length that is larger than some
UB could be used further and is not considered as
waste. The question is how to determine UB. The
answer depends on the quantity of available stock
lengths.
Let us see case 1 rst. If sucient stock lengths
are available then there will be cutting plans with
``no trim loss'' but ever growing stocks. To prevent
this an additional condition (case 1, condition (4))
has to be set: only one residual length may be
longer than the longest order length. UB can be set
arbitrarily between 0 and max si . The bigger UB
the greater the cutting problem. UB min si is
found in practice [5].
In case 2, however, UB max si . If, for example, UB is reduced to min si , this would lead to the
following problem: As the aim of the algorithm is

minimisation of the overall trim loss, this could


lead to unfullled requirements for the longest
order lengths, even if the overall trim loss is small
and the aim is achieved according to the logic of
the algorithm. The trim losses, which would be
longer than UB but shorter than the longest order
lengths, could remain unutilised. Because of this
setting UB in the case 2 is not reasonable and is
not included in the model.
3. Solution development
The proposed algorithm was developed on a
step by step basis. The number of basic steps
equals to the number of stock lengths necessary for
fulllment of an order. At the beginning, all stock
lengths belong to the set of unprocessed stock
lengths. The set of processed stock lengths is
empty. At each step, the set of unprocessed stock
lengths is reduced by one and the set of processed
stock length increases by one. Also, the number of
cut pieces of particular order lengths changes, as
well as the length of the processed stock length,
which becomes equal to trim loss. Algorithm has
the following steps:
Step 1: Select order lengths.
Step 2: Select stock length and cut it with chosen order lengths.
Step 3: If all stock lengths are not cut yet and
the requirements for order lengths are not fullled, then go back to step 1, else stop.
The algorithm is further developed on the following assumptions:
It is easier to nd a good solution if:
1. we can choose from the largest possible set of
possible solutions.
The number of possible solutions is higher if:
1.1. there are available as many as possible
order lengths that are not yet cut to the
end;
1.2. the proportion between average stock
lengths and average order lengths is as
large as possible;
1.3. the number of uncut order lengths is

M. Gradisar et al. / European Journal of Operational Research 114 (1999) 557568

as high as possible. In the best possible


case,
! !
m
X
xij  si
min
bi
j1

P max dj  1 zj :
2. the solutions dier from each other as much
as possible.
The dierences between possible solutions
are greatest when:
2.1. the relation between the longest and
the shortest order length is as great as possible;
2.2. the relation between the longest and
shortest stock length is as great as possible.
The assumptions are statistically proven, which
is shown in the analysis described in Section 4.
However, we cannot be quite sure that the assumptions are correct in each individual case.
Reduction in the inuence of ending conditions
requires such an algorithm, that at the end of the
cutting process it would be possible to choose from
the largest possible set of solutions diering from
each other as much as possible.
When there is enough material in stock, or
more than needed (case 1) all the assumptions
above can be taken into consideration more easily.
In this case is the last cut stock length mostly only
partially utilised and there is no additional trim
loss. The greater the number of stock lengths left
unutilised after the order requirements have been
fullled, the greater the possibility to achieve true
cutting optimum.
The crucial question is the selection of order
lengths. The following two procedures can be
considered:
(a) In accordance with 1.1 and 2.1 order lengths
with the greatest number of uncut pieces is chosen.
We choose rst Y elements from the set of not
completely cut order lengths, sorted by decreasing
number of uncut pieces (Di ). In this way a greater
variety of order lengths will remain at the end.
However, this is not consistent with the assumptions 1.2 and 1.3, as at the end of the cutting
process also longer order lengths could remain,
which would mean a smaller number of pieces to

561

be cut from the rest of the stock. This would be


appropriate only when the dierences between the
longest and shortest order lengths are small.
(b) According to assumptions 1.2 and 1.3 the
longest order lengths, or the rst Y elements in the
set of not yet entirely cut order lengths sorted by
decreasing lengths, are chosen. All this disagrees
with assumption 1.1, as only the shortest order
lengths remain, and with 2.1 as, because of this,
the dierence in their length is relatively small.
Therefore such a choice would be appropriate only
when the dierences between the longest and
shortest order lengths are great.
As assumptions 1.11.3, 2.1 and also 2.2 cannot
be met at the same time a compromise should be
considered. Assumption 2.2 is considered as less
important and is ignored. There are a few possible
solutions. A particular number of order lengths
can be chosen in one way and the rest in the other.
This number is marked f, where: 0 6 f 6 Y. Order
lengths can also be chosen on the basis of the total
length of not yet cut pieces. There are three possibilities:
1. f order lengths are chosen using procedure (a),
while Y ) f are chosen using procedure (b);
f 0,. . .,Y.
2. f order lengths are chosen using procedure (b),
and Y ) f using procedure (a); f 1,. . .,Y ) 1.
3. Order lengths can be chosen from the arrangement made according to the decreasing value of
si Di .
There are 2 Y + 1 possible solution variants. The
cutting algorithm is designed so that all variants
compete.
The issue of stock length selection can be
tackled by calculating the solutions for all unprocessed stock lengths and selecting among them the
one with the lowest trim loss. As it is possible that
more stock lengths have the lowest trim loss, we
apply assumption 1 and choose the shortest stock
length. This can be achieved with the initial arrangement of the stock lengths according to increasing lengths. By doing so, the longer stock
lengths will be processed later. Sometimes only one
best solution exists. Even if there are more of them
they are more likely at longer stock lengths. In this
case, at the beginning shorter stock lengths are
more preferred, which is achieved by increasing

562

M. Gradisar et al. / European Journal of Operational Research 114 (1999) 557568

parameter r. Parameter r is dened as a trim loss


that can be neglected. All the solutions with the
trim loss equal or smaller than r can be considered
as equally good. The algorithm for the optimisation of stock length cutting is shown with a owchart in Fig. 1.
If not enough material is available and a required distribution of uncut pieces by order
lengths is to be considered, the algorithm (Fig. 1)
applies to all distributions, yet it is not carried out
completely. In the case 2.2.1 (a) only procedure (a)
is carried out which means that f 0 and k 1. As
in every basic step only the lengths with the highest
b are chosen, it can be expected that at the end of
the procedure the shortage will be equally distributed through all order lengths. In case 2.2.1 (b)
only procedure (b) is carried out, which means
only when f 0 and k 0. As in every basic step
the longest order lengths are chosen it can be expected that only the shortest order lengths should
remain at the end. In case 2.2.1 (c) the algorithm
should be changed and order lengths sorted by
increasing lengths. However, this would be in
discrepancy with all four basic assumptions (1.1
1.3 and 2.1) and would cause a great trim loss.
Therefore this case is considered in the same way
as 2.2.2, where the problems are transformed in
the form described in case 1.
The owchart indicates that it is necessary to
solve a series of knapsack problems for each basic
step of algorithm. The dynamic programming
scheme of the procedure KNAPSACK in the
procedure CUT which selects and cuts the stock
length with the lowest trim loss Rmin according to
rst optimal combination of four (Y 4) selected
order lengths s1 , s2 , s3 , and s4 , can be summarized
as follows:

9.
R D 3 ) s4 t
10.
if R < Rmin
11.
then
12.
Rmin R
13.
jmin j
14.
x1j i, x2j k, x3j l, x4j t
15.
if R 6 r
16.
then
17.
stop
18.
endif
19.
endif
20.
endfor
21.
endfor
22.
endfor
23. endfor

Procedure KNAPSACK
0. initialize Rmin maxint
1. for j 1 to m do
2.
for i min{b1 , int(dj /s1 )} to 0 step-1 do
3.
D1 dj ) s1 i
4.
for k min{b2 , int(D1 /s2 )} to 0 step-1 do
5.
D2 D1 ) s2 k
6.
for l min{b3 , int(D2 /s3 )} to 0 step-1 do
7.
D3 D2 ) s3 l
8.
t min{b4 , int(D3 /s4 )}

O m

In the knapsack procedure a sequence of vectors (xj )Y is generated in a lexicographically decreasing order to nd the rst optimal
combination. The int function converts a numeric
expression to an integer, all digits to the right of
the decimal place are ignored.

3.1. Complexity analysis


Step 3 takes
Om  intdj =s1 ;

step 5 takes
O m

intd
j =s1
X

intdj s1  i=s2 ;

i0

and steps 719 take


intd
j =s1
X
i0
intdjX
s1 i=s2

!
intdj s1  i s2  k=s3

k0

time.
In case s1 s2 s3 s and Nj int(dj /s) step 3
takes O(m Nj ), step 5 takes Om  Nj2 3Nj ;
and steps 719 take Om  Nj3 6Nj2 11Nj
time.

M. Gradisar et al. / European Journal of Operational Research 114 (1999) 557568

Fig. 1. Flowchart of the cutting algorithm.

563

564

M. Gradisar et al. / European Journal of Operational Research 114 (1999) 557568

Time complexity depends on m, si and dj . The


procedure KNAPSACK itself depends on Y. For
Y 3 steps 6, 7, and 20 should be left out and time
complexity can be calculated with expression (1)
and (2). For Y 2 steps 47, 20, and 21 should be
left out and time complexity can be calculated with
expression (1). For Y 1 steps 27, 2022 should be
left out and time complexity is O(m). The greatest
time complexity is for Y 4 but for common values,
for example m 50 and Nj 30, the algorithm is
very fast and a short response time can be expected.

4. Results
The proposed algorithm is written in FORTRAN programming language. The program
consists of 2500 lines of code. The data input and
the printout of the results are made in 4GL, as it is
most suitable for this purpose. The program can
be run on a personal computer. It was called CUT.
CUT is intended for general use and is an improved and generalised version of program COLA
[5]. COLA was designed for cutting rolls in

Fig. 2. Two examples of the CUT program printout.

M. Gradisar et al. / European Journal of Operational Research 114 (1999) 557568

clothing industry and is in use for more than 12


years.
Creating a cutting plan for an extensive order
consisting of 50 stock lengths and 1000 pattern
pieces takes less than 20 s on a personal computer
(Pentium, 133 MHz). The time spent on creating
the plan is negligible. The speed enables the users
to carry out a ``what-if'' analysis by changing the
parameters of the order. For instance, a possible

565

question could refer as to what extent the solution


would be dierent if we change Y from 3 to 2 or r
from 0 to 5.
Two examples of computational results with
randomly generated data are seen in Fig. 2. The
customer orders are rather small, containing only
ve dierent order lengths and only four and ve
stock lengths. In the rst case we are short of
material so only 120 out of 125 anticipated pieces

Fig. 2. (continued).

566

M. Gradisar et al. / European Journal of Operational Research 114 (1999) 557568

can be carried out. The dierence is approximately


equally distributed according to order lengths.
Total trim loss is 2 cm, which makes 0.0046% of
total length. The saving of the material is 7.23 m.
In the second case there is no shortage of material.
The devised cutting plan assumes a complete cutting of four stock lengths and partial cutting of
one stock length. The anticipated trim loss is 0 cm.
The saving of the material in this order is 8.1 m. In
both cases creating a cutting plan takes less than
1 s.
The column ``the following order length'' (foll.
o.l.) contains order length reference numbers
which are about to be cut from the stock lengths
given in the column ``stock length''. This item of
information is helpful for the worker at the cutting
machine.
4.1. Parameters aecting utilisation of stock
With the program CUT basic assumptions 1
and 2 in Section 3 are statistically tested. Let us
dene two parameters p1 , and p2 p1 , is the average number of pieces cut out of each stock length.
The initial number of possible solutions depends
on p1 p2 is the average dierence between the
longest and the shortest order length. The initial
dierences between possible solutions are determined with parameters p2 and Y.
The inuence of p1 , p2 and Y has been analysed
by calculating a large number of random gener-

ated orders with dierent values of parameters in


accordance with the sample case and thus obtaining the average values of trim loss. The sample
case presents to the situation of cutting rolls in
clothing industry so that the results could be
compared with those obtained with the program
COLA. The sample case can be described as follows:
order lengths, number of pieces and stock
lengths are generated by sampling an integer
from the uniform distributions [300, 500], [10,
50] and [9000, 15 000],
Y 4, r 0,
material in excess is approximately 5%.
Let us consider rst the impact of parameter p1 .
If p1 is higher there is more room for manoeuvre
when seeking the optimal combination, and the
result can be better. Very good results can be
achieved if p1 is more than 30. Fig. 3 shows the
conditions.
Fig. 4 shows dependence of the trim loss on p2
expressed as a percentage. It can be observed that
p2 between 30% and 40% can suce for the optimal utilisation of material.
A crucial parameter, which is set as a parameter
of the program, is the number of dierent order
lengths, cut out of a stock length. A higher Y
means more possible combinations and, therefore,
also lower trim loss. Analysis shows that in case of
a combination of two order lengths in a stock
length the average trim loss accounts for 0.1143%.
In combinations of three the average trim loss is

Fig. 3. Dependence of the trim loss on the average number of pieces cut out of a stock length.

M. Gradisar et al. / European Journal of Operational Research 114 (1999) 557568

567

Fig. 4. Dependence of trim loss on dierences in order lengths.

Fig. 5. Dependence of trim loss on the number of dierent order lengths cut out of the same stock length.

reduced to 0.0121%. In combinations of four the


average trim loss is 0.0034%. It is noteworthy that
even when cutting only one order length out of a
stock length as in the case of the traditional procedure without a cutting plan, implementing the
CUT program can lead to the trim loss reductions
of up to 70% as shown in Fig. 5. This can be
achieved by merely placing the stock lengths to be
cut into a sequential arrangement proposed in the
program.
As we can see the values of all three parameters
p1 , p2 and Y of the sample case are great enough,
so a small trim loss can be expected. The question
is how great is the trim loss and what is the
probability of an optimal solution when all three
parameters have favourable values. In our case the

values are: p1 30, p2 40% and Y 4. We carried out an analysis of 200 randomly generated
samples of the sample case. In 100 cases the
shortage of material was about 5%, in other 100
cases there was about 5% surplus.
We obtained the following results: in the rst
100 cases the average trim loss was 0.0098%
(0.1121% in the worst case). In the other 100 cases
the average trim loss was 0.0034% (0.0351% in the
worst case). In all 200 cases the sum of all trim
losses was lower than the shortest order length.
This means that we have certainly found an optimal solution. Comparison with the program COLA shows an about two times smaller trim loss.
Two typical examples from these 200 cases are
shown in Fig. 2.

568

M. Gradisar et al. / European Journal of Operational Research 114 (1999) 557568

5. Conclusion
The paper analyses the problem of reducing
trim loss in one-dimensional stock cutting when all
stock lengths are dierent. The item-oriented solution in the form of an SHP through a combination of approximation and heuristics is found.
The algorithm is designed so that the conditions
and restrictions that could appear in practice are
taken into account. Such an algorithm is supposed
to be applicable as generally as possible in a variety of real life situations. On the basis of the
proposed algorithm the computer program CUT is
developed. This program is an improved and
generalised version of the program COLA, which
has been used in clothing industry for more than
12 years. The program is very fast. Its speed enables the users to carry out a ``what-if'' analysis by
changing the parameters. When all the parameters
are similar to or greater than those in the sample
case, the optimal solution can be expected in all
likelihood.
References
[1] J.S. Ferreira, M.A. Neves, P. Fonseca, A two-phase roll
cutting problem, European Journal of Operational Research 44 (1990) 185196.
[2] R.W. Haessler, P.E. Sweeney, Cutting stock problems and
solution procedures, European Journal of Operational
Research 54 (1991) 141150.

[3] R.W. Haessler, M.A. Vonderembse, A procedure for


solving the master slab cutting stock problem in the stell
industry, AIIE Transactions 11 (1979) 160165.
[4] E.E. Bisho, G. Waesher, Cutting and packing, European
Journal of Operational Research 84 (1995) 503505.
[5] M. Gradi, J. Jesenko, G. Resinovic, Optimization of roll
cutting in clothing industry, Computers & Operations
Research 24 (1997) 945953.
[6] H. Dyckho, A typology of cutting and packing problems,
European Journal of Operational Research 44 (1990) 145
159.
[7] P.C. Gilmore, R.E. Gomory, A linear programming
approach to the cutting stock problem, Operations
Research 9 (1961) 849859.
[8] P.C. Gilmore, R.E. Gomory, A linear programming
approach to the cutting stock problem, Part II, Operations
Research 11 (1963) 863888.
[9] E.G. Coman Jr., M.R. Garey, D.S. Johnson, Approximation algorithms for bin packing An updated survey,
in: G. Ausiello, M. Lucertini, P. Serani (Eds.), Algorithm
Design for Computer Systems Design, Springer, New
York, 1984, pp. 49106.
[10] P.E. Sweeney, E.R. Paternoster, Cutting and packing
problems: A categorised, application-orientated research
bibliography, Journal of the Operational Research Society
43 (1992) 691706.
[11] P.E. Sweeney, R.W. Haessler, One-dimensional cutting
stock decisions for rolls with multiple quality grades,
European Journal of Operational Research 44 (1990) 224
231.
[12] G. Waesher, T. Gau, Generating almost optimal solutions
for the integer one-dimensional cutting stock problem,
Working Paper No. 94/06, Institut f
ur Wirtschaftswissenschaften, Technische Universitat Braunschweig,
1994.

Você também pode gostar