A tabu search algorithm

for the pallet loading problem

R. Alvarez-Valdes † , F. Parreño ‡ , J.M. Tamarit † ,

† University of Valencia, Department of Statistics and

Operations Research, 46100 Burjassot, Valencia, Spain
‡ University of Castilla-La Mancha. Departamento de

Informatica, E.Politecnica Superior, 02071 Albacete, Spain

Abstract
This paper presents a new heuristic algorithm for the pallet loading
problem, the problem of packing the maximum number of identical
rectangular boxes onto a rectangular pallet. The problem arises in
distribution and logistics and has many practical applications. We
have developed a tabu search algorithm based on new types of moves.
Instead of moving individual boxes, we propose moving blocks, sets of
boxes with the same orientation. We have tested our algorithm on
the whole sets Cover I and Cover II, usually taken as a reference for
this problem, and we obtain excellent results in very short computing
times.

1 Introduction
In the Pallet Loading Problem, PLP, the maximum number of identical rect-
angular boxes have to be packed onto a rectangular pallet. The problem
arises at factories when large quantities of one product must be shipped
onto pallets, and it is sometimes called the Manufacturer’s Loading Problem
(MLP) to distinguish it from the Distributor’s Loading Problem (DLP) in
which in boxes of several sizes are packed together on one pallet. Though
the problem is initially three-dimensional, practical considerations of stabil-
ity and handling usually mean that the boxes must be placed orthogonally
with respect to the edges of the pallet and in the layers in which the ver-
tical orientation of the boxes is fixed. With these restrictions the problem
becomes the two-dimensional problem of packing a large rectangle, a pallet,
with the maximum number of small identical rectangles, boxes.

1
 The problem has many practical applications in distribution and logistics.
An increase in the number of boxes which can be shipped on a pallet directly
leads to a decrease in costs. Though the problem is very easy to solve in
many cases, it cannot be considered to have been completely solved. For
example, for the instance (89,75,10,7), a state-of-the-art commercial code for
integer programming such as CPLEX took 89 hours of CPU time to obtain
an optimal solution with 99 boxes. The problem has therefore continued to
attract a lot of research in recent decades.
The exact algorithms proposed to date are basically tree search proce-
dures in which at each node a partial layout of boxes on the pallet has been
built. Different ways of adding boxes, extending the partial solution, and dif-
ferent bounding procedures define the different algorithms [2],[5],[8],[9], [11],
[15]. These exact algorithms have been able to solve problems of only mod-
erate size, up to 50 boxes. Several upper bounds have also been proposed
[6],[7],[16],[19], which consider the geometric structure of the problem and
linearly relax integer programming formulations.
Many heuristic methods have been developed. On the one hand, there
are constructive methods of increasing complexity, from simple structures in
which the pallet is divided into blocks [3],[21], [22], to recursive procedures
[17],[18], [20],[23]. On the other hand, there are metaheuristics based on tabu
search, genetic algorithms and strategic oscillation [1],[10],[13].
The best results to date have been obtained by recursive procedures
that exhaustively study special structures leading to good but not neces-
sarily optimal solutions. The algorithm G − 4 of Scheithauer and Terno [20]
finds, in very short computing times, the best G4-structures of each prob-
lem, structures which produce the optimal solution for all the instances of
set CoverI (up to 50 boxes) and for most of the instances of set CoverII
(up to 100 boxes). The algorithm proposed by Lins et al.[17] finds the best
L-decomposition. This structure contains G − 4 as a special case and obtains
more optimal solutions, but it is computationally much more expensive. The
metaheuristic algorithms described so far present computational results lim-
ited to small subsets of problems and they have not yet obtained results
comparable to those sophisticated recursive procedures.
The purpose of the present paper is to develop an algorithm, based on
tabu search, that can compete with the best recursive procedures in terms of
the quality of the solutions and the computing times. The structure of the
paper is as follows. In Section 2, we formally define the problem and review
the main concepts, especially the idea of equivalence classes. In Section 3

2
we present the main components of the algorithm, defining the moves, the
tabu list and the intensification and diversification strategies. In Section 4 we
present the computational results. Finally, concluding comments are made
in Section 5.

2 Problem formulation and basic concepts

The problem can be formulated as follows: Given a large rectangle, a pallet,
with given length L and width W , and a small rectangle, a box, with length a
and width b, the number of a ∗ b boxes packed onto the L ∗ W pallet has to be
maximized. Each instance of the PLP is denoted by a quadruple (L, W, a, b).
Positions (x, y) of the boxes are given with respect to a coordinate system
which originates in the bottom left corner of the pallet.
Initially each box can be placed at any position on the pallet, that is,
x ∈ X = {u|0 ≤ u ≤ L − b, integer}, y ∈ Y = {v|0 ≤ v ≤ W − b, integer}.
The set of positions for the boxes can be first reduced to the normal sets
proposed by Herz[14] and Christofides and Whitlock[4]:
S(L) = S(L, a, b) = {r : r = αa + βb, r + b ≤ L, α, β ∈ Z+ }
S(W ) = S(W, a, b) = {r : r = αa + βb, r + b ≤ W, α, β ∈ Z+ }
These sets can be further reduced by using dominance considerations, as
proposed by Scheithauer and Terno[20]. If we denote:
hsiL := max{r ∈ S(L) : r ≤ s}
the new set of positions, raster points, are:
e
S(L) e a, b) = {hL − ri : r ∈ S(L)}
= S(L, L
e ) = S(W,
S(W e a, b) = {hW − riW : r ∈ S(W )}
Given an instance (L, W, a, b), a pair (n, m) of non-negative integers is a
feasible partition of S (L or W ) if n ∗ a + m ∗ b ≤ S.
If (n, m) satisfy 0 ≤ S − n ∗ a − m ∗ b < b, they are an efficient partition.
If (n, m) satisfy 0 = S − n ∗ a − m ∗ b, they are a perfect partition.
Dowsland[6] showed that two instances of the PLP are equivalent, that is,
have the same feasible solutions, if they have the same efficient partitions for
both the length and the width of the pallet. Hence, if we solve one instance
of an equivalence class, in particular that with the lowest dimensions, we
have solved all the other instances of the class.
The concept of equivalence class also determines the sets of test instances
used for the PLP. Randomly generating and solving a set of instances would

3
suppose solving equivalent problems more than once. It seems more rea-
sonable to work with equivalence classes. Two sets have been proposed[9]:
Cover I, the set of equivalence classes of instances satisfying:
L a L∗W
1≤ ≤ 2, 1≤ ≤ 4, 1≤ < 51
W b a∗b
and Cover II, the set of equivalence classes of instances satisfying:
L a L∗W
1≤ ≤ 2, 1≤ ≤ 4, 51 ≤ < 101
W b a∗b
The definition of sets Cover I and Cover II is subject to some ambiguity.
Sometimes one instance satisfies the conditions defining the set but another
equivalent instance does not. In these cases, the inclusion of the equivalence
class in the set is not clear. In order to avoid this ambiguity, we have fol-
lowed the constructive method proposed by Dowsland[9] to generate all the
equivalence classes for which the lowest dimensional instance strictly satisfies
the defining conditions of the set. Cover I produced in this way has 7827
classes, and Cover II has 40609 classes.
The G-4 structure was defined by Scheithauer y Terno[20] and it is a very
efficient packing structure for the PLP. Some examples appear in Figures 6,
7, 8(a). Most of the optimal solutions for the problem contain some type of
G-4 structure.

3 The Tabu Search algorithm

Tabu Search is now a well-established optimization algorithm (for an intro-
duction, refer to the book by Glover and Laguna[12]. The basic elements of
the algorithm are described in the next subsections.

3.1 Definition of a move

The solution space in which we move is composed of feasible solutions only.
In this space we will define several moves to go from one solution to another.
Analyzing different solutions for the problem, we observe that, rather than
individual boxes, there are blocks, sets of boxes with a common orientation.
Boxes not included in a block appear very seldom, unless they are included
in a G-4 structure. Therefore, the moves we use are based on blocks and G-4

4
structures. This is a completely different approach from those already used
in other metaheuristic algorithms.
The move will consist of changing the size of a block of the solution. We
distinguish two types of moves:

• Block reduction
The size of a block is reduced horizontally or vertically by eliminating
some of its rows or columns. That includes the possibility that the
block may disappear completely.
The move starts by eliminating some of the rows or columns of the se-
lected block from the current solution, then resulting in the appearance
of a new region of waste, space not occupied by boxes. We move then
the remaining blocks towards the nearest corners of the pallet so that
waste is located mostly towards the center.
In order to reutilize the new waste, the first step is to merge the waste,
putting together the existing waste rectangles, new and old, in such a
way that the new space can be filled by the maximum number of boxes.
The waste rectangles are then filled, considering the horizontal or ver-
tical block or the G4-structure that best fills each of them . Finally,
adjacent blocks with the same structure are merged to simplify the so-
lution. Examples of reduction moves appear in Figures 1 and 2. In
Figure 1 a block is selected and reduced. None of the blocks is moved
to the corners, the wastes are merged and then filled in two steps to
produce a solution with more boxes. In Figure 2 the selected block dis-
appears, another block is moved to the bottom right corner, the waste
is merged and filled thereby producing a new solution which is optimal.

• Block augmentation The size of a block is increased horizontally or

vertically by adding some rows or columns. The enlarged block may
intersect with other blocks in which case the overlapping boxes are
eliminated from the other blocks.
Once we have enlarged the block and eliminated overlapping, we merge
the waste, then fill the waste rectangles and finally adjacent blocks with
the same structure are merged to simplify the solution, following the
same steps of block reduction. Some examples of block augmentation
appear in Figure 3.

5
 (a) Selection (b) Reduction (c) Filling

Figure 1: Block reduction. Instance (42,31,9,4)

(a) Selection (b) Reduc- (c) Moving to (d) Filling

tion the corner

Figure 2: Block reduction. Instance (38, 36, 10, 3)

(a) Selection (b) Increase (c) Elim. overlap.

Figure 3: Block augmentation. Instance (24,14,5,3)

6
We now describe the procedures used in the moves:

1. Merging waste rectangles

We need a procedure to merge the waste rectangles which will be ap-
pearing in the process. This procedure is essential for the quality of
the move. When we merge 2 rectangles, at most 3 new rectangles may
appear, usually one large rectangle and 2 small ones (see Figure 4).
Among the several alternatives of merging we try to select the most
promising one. With this objective in mind, we impose some condi-
tions. First, the rectangles to be merged must be adjacent and the
common side must have a length of at least b to accommodate at least
one box. Second, after the merge, the new rectangles must accommo-
date more boxes than before the merge.

Figure 4: Merging 2 waste rectangles

2. Moving blocks to the nearest corners

Every time a block is subject to some modification, we try to move it
to its nearest corner, so that waste is concentrated in the middle, where
it can be more easily merged (see Figure 5).

7
 Figure 5: Moving blocks towards the corners

3. Filling waste rectangles

We need a fast procedure to study efficient ways of filling each element
of the list of waste rectangles. For each rectangle, we consider three
possibilities: a horizontal block, a vertical block or a G4-structure and
take the one accommodating the largest number of boxes. If there is
a tie, we use blocks rather than G4-structures. If the waste has been
produced by the reduction of a horizontal block, we prefer to fill it with
a vertical block and conversely.
The procedure works iteratively. The rectangle may be partially filled
and the unfilled part is a new waste rectangle which is considered again
for filling until its surface is smaller than that of a box.
As was mentioned above, the G4-structure introduced by Scheithauer
and Terno [20] is a very efficient packing structure for this problem and
it appears very often in good solutions for the problem. We therefore
consider it as an alternative to fill the rectangles. We do not study
all the possible G4-structures but only what we call the natural G4-
structure, that with the same number of horizontal and vertical boxes
and producing the minimum inner waste. For example, for a 7 ∗ 3 box,
the natural G4-structure will be that with 2 horizontal and 2 vertical
boxes, as in Figure 6(a), leaving an inner waste of 1 ∗ 1. However, there
may be alternative structures, maybe not so efficient in terms of the
inner waste, but fitting on the pallet better. In the example, a new
structure could be considered (Figure 6(b)), producing an inner waste
of 2 ∗ 2. We leave the study of these alternatives to the diversification
strategies described later.
When choosing the natural G4-structure we also take into account
whether its horizontal and vertical boxes would belong to perfect parti-

8
 Figure 6: Natural G4 structures for a 7 ∗ 3 box

tions of the length and width of the pallet. If this is not the case, we do
not choose the G4-structure. Proceeding in that way, it is possible that
no G4-structure with the same number of horizontal and vertical boxes
is left to be chosen. In this case, we consider structures with different
numbers of horizontal and vertical boxes, as appears in Figure 7.

(a) (b) (36,28,10,3) (c) (58,31,9,4)

(26,18,7,2)

Figure 7: Other types of natural G4-structures

Once we have chosen a G4-structure, we can obtain from it some other

derived structures, repeating them or nesting them (Figures 8(a),8(b)).

4. Merging blocks with the same structure

The number of moves to explore increases with the number of blocks.
Therefore, at the end of each iteration we try to merge blocks with the
same orientation and the same length or width if they are adjacent and
have a common side, or if one of them can be moved to make them
both adjacent to one common side (Figure 9).

9
 (a) (b)

Figure 8: Repeated and nested G4-structures

(a) Before merging (b) Merged block

Figure 9: Merging blocks

10
3.2 Reducing the number of moves to explore
If we have a block of dimension (n × m), we can do (n + m − 1) reductions
and, depending on the size of the pallet, a number of augmentations. From
all these possible moves, we are not going to study those considered not
promising, according to several criteria. In that way, we increase the speed
of the procedure, hopefully without a significant decrease in the quality of
the solutions.

• If a block is not adjacent to any waste rectangle, it is not considered

for moving. If the block is adjacent to a waste rectangle, we consider
moves only in the direction in which there is waste.

• When we reduce or enlarge a block, we only explore the move if the new
block can form part of an efficient partition of the length and width of
the pallet.

• If the bottom left corner of the new block is not a rater point, we do
not consider the move.

• If the new block will produce more waste than that associated with the
best known solution, we do not explore the move.

3.3 Moves to be studied

At each iteration we apply the following three-step procedure to each block
of the current solution:

1. Select a block adjacent to a waste rectangle.

2. Consider all types of moves, reductions and augmentations, in the di-

rections adjacent to waste rectangles.

3. For each type of move, consider all possible numbers of rows and
columns not eliminated by the criteria described above.

3.4 Selection of the move

The objective function consists only of maximizing the number of boxes
packed onto the pallet. However, if the moves are evaluated according to that

11
function, there will be many moves with the same evaluation. Therefore, we
have included other desirable characteristics of the solution in the evaluation
function, which are listed below ordered in decreasing importance:

• Symmetry. We try not to explore symmetric solutions but prefer solu-

tions in which waste is mostly concentrated to the right and to the top
of the pallet.

• Efficiency. If possible, we prefer solutions in which the blocks are

efficient. We say that a block is efficient if its length and width can
make part of a perfect partition.

• Centered and concentrated wastes. We prefer solutions in which waste

rectangles are centered and concentrated as much as possible, because
that will make it easier to merge them and obtain spaces for more
boxes.

• Waste in pallet sides. Solutions without waste in the sides of the pallet
are preferred.

• Improving. We do not allow solutions with a block which cannot appear

in a solution which is better than the best known solution.

• New boxes. We prefer a solution with some new boxes, rather than a
solution which is the same as the previous solution but eliminates some
of its boxes.
These criteria are added to the evaluation function with some weights
reflecting their relative importance, according to the results of a pre-
liminary computational experience on a subset of problems.

3.5 Tabu list

At each iteration the block selected for the move is added to the tabu list. A
move is then tabu if the new solution contains a block in the tabu list. For a
given number of iterations we cannot visit a solution with a block with the
same size and at the same position of a selected block.
The tabu list size varies dynamically. After a given number of itera-
tions without improving the solution, the length is randomly chosen from
[0.5m∗ , 1.5m∗ ], where m∗ =⌊ L∗W
l∗w
⌋ (the upper bound based on the pallet area).

12
 The aspiration criterion allows us to move to a tabu solution if it improves
the best solution obtained so far.

3.6 Local improvement

At the end of each iteration we try a local improvement. We put a G-4 struc-
ture at every corner of the pallet, one at a time, eliminate the overlapping
boxes, and see if the new solution is better than the current solution.

3.7 Intensification and diversification strategies

The moves we have defined involve a high level of diversification. However,
we have included two more diversification strategies:

• Strategic oscillation
Our evaluation function has several terms whose relative importance
is controlled by a set of weights. These weights have been chosen for
having a good average behavior, but maybe they are not the most
appropriate for some particular instances. Therefore, we make them
oscillate in the following way:
We use a coefficient βi , i = 1, . . . 6 which multiplies every weight. Ini-
tially every βi = 1, i = 1, . . . 6. After a given number of iterations
without improving the best known solution, the coefficient oscillates
between 0 and 1.

• Long term memory

Along the search process, we keep in memory the frequency of each
block appearing in the solutions, where a block is defined by its size
and orientation.
This information is used for diversification and for intensification pur-
poses. When used for diversification, we favor the moves of blocks
appearing very frequently in the solutions, then inducing new blocks to
appear. When used for intensification, we keep in memory only blocks
corresponding to high quality solutions and then we favor these blocks
appearing again in the new solutions.

13
4 Computational results
We have coded our algorithm in C++ and made our tests on a Pentium 4
at 2.0GHz. After every 100 iterations without improving the best known
solution we change the natural G-4 structure to be considered, if there is
more than one alternative. Also, the length of the tabu list changes every 100
iterations without improvement. After 10000 iterations without improving
we perform a strategic oscillation of the objective function coefficients and
after 9000 non-improving iterations we do a diversification phase based on
long term frequencies over 250 iterations and then an intensification phase
over 250 iterations.
All the 7827 classes of Cover I were solved to optimality in less than 250
iterations. That supposes less than 1 second of CPU time. The results for
the complete set Cover II, 40609 classes, appear in Table 4, which shows the
number of instances not optimally solved. A limit of 100000 iterations has
been imposed in order to keep the CPU time below the 10 minutes.

Table 1: Cover II (40609 classes).

Iterations Non-optimal
500 229
1000 137
10000 38
50000 23
100000 15

Heuristic G-4 of Scheithauer and Terno [20] is available on Internet and

we have used it to solve all the classes of Cover II. Only 45 of them were not
solved to optimality because they do no not have a G-4 structure. There-
fore, our Tabu Search algorithm, which solves all the instances solved by
the G4 procedure and some others, is prepared to go one step further and
find optimal solutions for problems not having G4-structure at a reasonable
computational cost. Figure 10 shows an example of optimal solution with no
G-4 structure.
The algorithm proposed more recently by Lins et al. [17] is more com-
plex. It exhaustively studies L-structures, which include G-4 structures as
particular cases. For simplicity the authors have restricted their study to
a subset of Cover II, of more than 20000 instances, and it has solved all

14
 Figure 10: Instance (104,69,12,7) of Cover II

of them optimally, though in some cases the computational times are very
high (more than 360 minutes on a Pentium III at 700 Mhz). In fact, they
conjecture that their algorithm always finds optimal solutions for the pallet
loading problem. If this conjecture could be proven, the algorithm would no
longer be heuristic and its lengthy computational times should be compared
with other exact approaches.
On this subset of instances our Tabu Search algorithm has found the
optimal solution for all but one instance, (74,46,7,5), within the imposed
limit of 10 CPU minutes. Therefore, our algorithm seems to achieve a good
balance between solution quality and computational cost.
Finally, in order to explore the possible extension of our algorithm to
larger problems, we have generated what we have called Cover III, the set of
classes satisfying:
L a L∗W
1≤ ≤ 2, 1≤ ≤ 4, 101 ≤ < 151
W b a∗b
extending the structure of Cover I and Cover II to instances of up to 150
boxes. There are 98016 equivalence classes for which the optimal solution is
not known, but for which several upper bounds can be computed. For that
set, we are able to compare the solutions obtained by our heuristic and the
upper bounds. Table 4 shows the number of problems for which the solution
obtained by the heuristic does not match the corresponding upper bound.
These results allow us to conclude that the tabu search algorithm optimally

15
solves most of the classes of Cover III. Figure 11 shows a solution with 149
boxes.

Table 2: Cover III (98016 Instances).

Iteration Non-optimal
1000 1770
10000 1263
25000 1202
50000 1173
100000 1122

Figure 11: Instance (153, 88, 18, 5) of Cover III

5 Conclusions
We have developed a new Tabu Search algorithm which produces excellent
results for the whole sets Cover I and Cover II and which can be extended
to a new set Cover III of equivalence classes for problem instances of up to
150 boxes.
New types of moves have been defined. On the one hand, increasing and
decreasing blocks and on the other hand those which explicitly consider the
inclusion of G4-structures which have been shown to be very efficient for this
particular packing problem. These moves are embedded in an algorithmic

16
scheme combining diversification and intensification strategies. The results
show that these metaheuristic procedures can obtain good results in reason-
able computing times for large-size pallet problems. We are currently trying
to apply this type of ideas to more complex packing problems, involving
boxes of different sizes.

References
[1] Amaral A, Wright M (2001) Experiments with a strategic oscillation algo-
rithm for the pallet loading problem. International Journal of Production
Research 39-11: 2341-2351

[2] Bhattacharya R, Bhattacharya S (1998) An exact depth-first algorithm

for the pallet loading problem. European Journal of Operational Research
110: 610-625

[3] Bischoff EE, Dowsland WB (1982) An Application of the Micro to Prod-

uct Design and Distribution. Journal of the Operational Research Society
33: 271-280

[4] Christofides N, Whitlock C (1977) An Algorithm for Two-Dimensional

Cutting Problems. Operations Research 25: 30-44

[5] De Cani P (1979) Packing problems in theory and practice. Ph.D.Thesis.

Department of Engineering Production, University of Birmingham.

[6] Dowsland KA (1984) The three-dimensional pallet chart: an analysis of

the factors affecting the set of feasible layouts for a class of two-dimensional
packing problems. Journal of the Operational Research Society 35: 895-905

[7] Dowsland KA (1985) Determining an upper bound for a class of rectan-

gular packing problems. Computers and Operations Research 12: 201-205

[8] Dowsland KA (1987) An exact algorithm for the pallet loading problem.
European Journal of Operational Research 31: 78-84

[9] Dowsland KA (1987) A combinated data-base and algoritmic approach

to the pallet-loading problem. European Journal of Operational Research
38: 341-345

17
[10] Dowsland KA (1996) Simple tabu thresholding and the pallet loading
problem. In: Osman IH, Kelly JP(Eds) Metaheuristics: theory and appli-
cations, Kluwer Academic Publishers, pp 379-406

[11] Exeler H (1988) Das homogene Packproblem in der betrieb-

swirtschaftlichen Logistik. Physica-Verlag, Heidelberg.

[12] Glover F, Laguna M (1997) Tabu Search, Kluwer Academic Publishers,

Boston.

[13] Herbert A, Dowsland W (1996) A family of genetic algorithms for the

pallet loading problem. Annals of Operations Research 63: 415-436

[14] Herz JC (1972) A recursive computational procedure for two-

dimensional stock cutting. IBM Journal of Research Development 16: 462-
469

[15] Iserman H (1987) Ein Planungsssytem zur Optimierung der Palettenbe-

ladung kongruenten rechteckigen Versangebinden. OR Spektrum 9: 235-
249

[16] Letchford A, Amaral A (2001) Analysis of upper bounds for the Pallet
Loading Problem. European Journal of Operational Research 132: 582-593

[17] Lins L, Lins S, Morabito R (2003) An L-approach for packing (l,w)-

rectangles into rectangular and L-shaped pieces. Journal of the Operational
Research Society 54: 777-789

[18] Morabito R, Morales A (1998) A simple and effective recursive procedure

for the manufacturer’s pallet loading problem. Journal of the Operational
Research Society 49: 819-828

[19] Neliβen J (1995) How to use structural constraints to compute an upper

bound for the pallet loading problem. European Journal of Operational
Research 84: 662-680

[20] Scheithauer G, Terno J (1996) The G4-Heuristic for the Pallet Loading
Problem. European Journal of Operational Research 46: 511-522

[21] Smith, De Cani P (1980) An algorithm to optimize the layout of boxes

in pallets. Journal of the Operational Research Society 31: 573-578

18
[22] Steudel JH (1979) Generating pallet loading patterns: A special Case
of the two-Dimensional Cutting Stock problem. Management Science 25:
997-1004

[23] Young-Gun G, Maing-Kyu K (2001) A fast algorithm for two-

dimensional pallet loading problems of large size. European Journal of
Operational Research 134: 193-202

19