Computational and Applied Mathematics (2022) 41:119

https://doi.org/10.1007/s40314-022-01825-2

New mixed-integer linear programming model for solving

the multidimensional multi-way number partitioning
problem

Bojan Nikolic1 · Marko Djukanovic1,2 · Dragan Matic1

Received: 15 January 2021 / Revised: 4 February 2022 / Accepted: 8 March 2022 /

Published online: 26 March 2022
© The Author(s) under exclusive licence to Sociedade Brasileira de Matemática Aplicada e Computacional 2022

Abstract
The multidimensional multi-way number partitioning problem considers a set of n vectors
of a fixed dimension m ≥ 2 as an input. The goal is to find partitioning of the set into k ≥ 2
subsets such that sums per each coordinate in each subset are as nearly equal as possible.
The problem has applications in key encryption and multiprocessor scheduling. Prior to this
paper, there has been only one mixed-integer model proposed in case of general k ≥ 2. In
this paper, we propose a new mixed-integer programming model for the general case. The
obtained results indicate that our model significantly outperforms the model from literature
for larger values of k.

Keywords Mixed integer linear programming · Partitioning problems · Mathematical

modelling

Mathematics Subject Classification 93A30 · 90C11

1 Introduction

Let S = {v1 , v2 , . . . , vn } be a set of n vectors, where each vector has m real coordinates. For
a fixed integer k ∈ {2, . . . , n}, the multidimensional multi-way number partitioning problem

Communicated by Joerg Fliege.

B Marko Djukanovic
[email protected]; [email protected]
Bojan Nikolic
[email protected]
Dragan Matic
[email protected]
1 Faculty of Natural Sciences and Mathematics, University of Banja Luka, Banja Luka, Bosnia and
Herzegovina
2 Institute of Logic and Computation, TU Wien, Austria

123
119 Page 2 of 72 B. Nikolic et al.

(MDMWNPP) asks for a disjoint and collectively exhaustive partition (S1 , S2 , . . . , Sk ) of set
S, where sums per each coordinate in each partition set are as nearly equal as possible.
MDMWNPP is a generalization of the well-known two-way number partitioning problem
(TWNPP) whose goal is to divide a set S of real numbers into two subsets, such that sums of
the numbers in each subset are as close as possible. MDMWNPP generalizes TWNPP in two
ways: (i) in MDMWNPP, instead of real numbers, the set S contains general m-dimensional
vectors and (ii) the number of subsets, when partitioning the set of vectors S, can be arbitrary
fixed integer k ≥ 2. TWNPP is one of Karp’s 21 basic N P –hard problems Karp (1975) (it
is also N P complete). Thus, MDMWNPP is N P –hard in a general case, as a generalization
of an N P –hard problem.
An important characteristic of TWNPP is that its computational complexity does not
depend (only) on the cardinality of set S, but on the type of input numbers, i.e., the number
precision Hayes (2002), Mertens (2006). As it is stated in Kojić (2010), this is not the case
with the multidimensional version of TWNPP; even in the case when k = 2, the problem is
hard to solve with a moderate number of digits in the input. The MDMWNPP in the case of
m = 1 is considered in Korf (2009), Moffitt (2013), Korf et al. (2014).
The first generalization which introduces vectors instead of numbers was given by Kojić
(2010). Kojić (2010) presented a mixed-integer linear programming (MILP) formulation of
this specific problem; this MILP formulation has n binary variables and 2m constraints. The
problem is mainly solved by various heuristic approaches such as a genetic algorithm (GA) by
Pop and Matei (2013), a variable and neighborhood search (VNS) and an electromagnetism-
like (EM) by Kratica et al. (2014), a greedy randomized adaptive search (GRASP) with
exterior path-relinking and restricted local search by Rodriguez et al. (2017). Moreover,
the following four approaches: GA, Simulated annealing (SA), migrating bird optimization
algorithm (MBO) and clonal selection algorithms are proposed by Hacibeyoglu et al. (2018),
and an Algebraic Differential Evolution and VNS by Santucci et al. (2019).
Bi-dimensional two-way number partitioning problem has been solved by greedy and
genetic algorithms by Hacibeyoglu et al. (2014). The same bi-dimensional problem is also
solved by Fuksz et al. (2013) by means of three heuristics, greedy, a novel GA and a hybrid of
GA and VNS method. The generalized MDMWNPP with arbitrary m and k is introduced by
Pop and Matei (2013). In a recent work, Faria et al. (2020) in [2] introduced the first known
exact algorithm, a mixed-integer programming model, to solve the generalized MDMWNPP.
The model has been tested for a smaller number of subsets (k ∈ {2, 3, 4}). For k = 2, the
model has shown as competitive to the model of Kojić (2010). More details about these two
models will be discussed in Sects. 2.1 and 2.2.
TWNPP and its generalizations face with many practical applications in areas like public-
key encryption and task scheduling. Practical applications of the NPP include multiprocessor
scheduling Coffman and Lueker (1991), public-key cryptography Merkle and Hellman
(1978), and the assignment of tasks in low-power application-specific integrated circuits
Ercegovac and Potkonjak (1999).
Regarding the generalized problems, MDTWNPP and the MDMWNPP are of a practical
usage when it is necessary to find a balance between two or more characteristics of elements
from a given data set. These applications include public-key cryptography Pop and Matei
(2013), and assigning persons to fair teams Hacibeyoglu et al. (2018).
Some concrete examples of the MDTWNPP may concern the following:

• Two travel agents need to have as closely similar revenues from organized tours as possi-
ble; the parameters under consideration are the price of the tour, the mileage (due to petrol

123
New mixed-integer linear programming model... Page 3 of 72 119

consumption), special offers (such as restaurants, shopping tours, visits to museums, and
so on). This is a three-dimensional, two-way number partitioning problem.
• Two police patrols need to have similar mileage, degree of risk (depending on criminal
activities) and time for touring their beat.

Beside further generalizations of the abovementioned examples, some concrete examples for
the MDMWNPP may be as follows:

• Cities in a country should be organized in regions, to achieve an equilibrium between the

obtained regions in terms of the observed characteristics, such as population, economic
power and area.
• A group of scientists wants to form more than one commission, so that, in terms of
competencies in the proposed areas, the formed commissions are of uniform quality.
• A professor wants to arrange tests for k students. He disposes with n questions and
each question is described by m measurable parameters, such as estimated difficulty and
estimated time needed to write the answer. Professor wants to deliver all questions to the
students, where each question is assigned to exactly one student and with the requirement
that both the overall difficulty and total needed time for each student is as balanced as
possible.

The main contributions of our paper are as follows: (i) we present a new mixed-integer
linear programming model to solve the MDMWNPP with a fixed number of subsets k ≥ 2;
(ii) we give an extensive computational studies also with larger k and compare our model to
the model of Faria [2] and (iii) our model has been competitive with this model for smaller
k, and is able to significantly outperform it significantly for a larger number of subsets k.

2 Problem definition and the MILP models from the literature

Let m ≥ 2 be a fixed integer and let S = {vi = (vi1 , . . . , vim ) ∈ Rm ) | i = 1, . . . , n} be a set

of n ≥ 2 vectors. For a fixed k ≥ 2, apairwise disjoint family {S1 , . . . , Sk } with the property
S j ⊆ S for each j = 1, . . . , k, and kj=1 S j = S is called a k–decomposition of the set S.
A k–decomposition with additional property that each of its elements is a non-empty set is
called a k–partition.
The multidimensional multi-way number partitioning problem (MDMWNPP) asks for
finding a k–decomposition of S such that the sums per every coordinate of elements in each
of its sets are equal or almost equal.
More precisely, for a given k–decomposition S = {S j | j ∈ {1, 2, ..., k}}, we define
⎛ ⎛ ⎞ ⎛ ⎞⎞
 
   
r (S ) := max ⎝  max ⎝ ⎠
vil − min ⎝ vil ⎠ .
⎠ (1)
l∈{1,...,m}  j∈{1,...,k} j∈{1,...,k} 
vi ∈S j vi ∈S j

The objective of the problem is to minimize r over all possible k–decompositions of S. Recall
that for k = 2, we get the MDTWNPP.

2.1 The model for the MDTWNPP from the literature

In this section, we present the MILP model proposed by Kojić (2010). The model is defined
as follows.

123
119 Page 4 of 72 B. Nikolic et al.

min r
s.t.

n
− 0, 5 · r + vil · xi ≤ 0, 5 · sl , ∀l ∈ {1, . . . , m} (2)
i=1

n
0, 5 · r + vil · xi ≥ 0, 5 · sl , ∀l ∈ {1, . . . , m} (3)
i=1
xi ∈ {0, 1}, ∀i ∈ {1, 2, . . . , n} (4)


n
where sl = vil , ∀l = 1, 2, . . . , m, and xi = 1, if vi ∈ S1 , or 0 other-
i=1
wise. This model has 2 m constraints and n variables. As one could notice, this model
allows empty subsets. For example, the optimal solution of the MDTWNPP for S =
{(−4, −4), (1, 1), (1, 1), (1, 1), (1, 1)} can be obtained from subsets

S1 = {(−4, −4), (1, 1), (1, 1), (1, 1), (1, 1)},
S2 = ∅.

2.2 The model for the MDMWNPP from the literature

This model has been proposed by Faria et al. (2020) in [2], and until now, it is the only known
method to solve the MDMWNPP. Apart of the model proposed by Kojić (2010), Faria et al.
(2020) considered only k–partitions of the set S. The model has mk(k −1)+k +n constraints
and nk + 1 variables.
Variables are (except of r ) given by xi j = 1, if vector vi is included in the subset S j ,
i ∈ {1, . . . , n}, j ∈ {1, . . . , k}, or 0 otherwise.

min r (5)
s.t.

k
xi j = 1, ∀i ∈ {1, . . . , n} (6)
j=1

n
xi j ≥ 1, ∀ j ∈ {1, . . . , k} (7)
i=1
n
vil (xi j1 − xi j2 ) ≤ r , 1 ≤ j1 < j2 ≤ k, ∀l ∈ {1, . . . , m} (8)
i=1
n
vil (xi j2 − xi j1 ) ≤ r , 1 ≤ j1 < j2 ≤ k, ∀l ∈ {1, . . . , m} (9)
i=1
r ∈ [0, +∞) (10)
xi j ∈ {0, 1}, ∀i ∈ {1, . . . , n}, ∀ j ∈ {1, . . . , k} (11)

Constraint (6) ensures that each element belongs to exactly one subset, and constraint (7)
ensures that each subset is non-empty. Constraints (8) and (9) ensures that for each pair of

123
New mixed-integer linear programming model... Page 5 of 72 119

different subsets, the absolute difference between these two subset sums per each coordinate
is at most r .

3 New model to solve the MDMWNPP

In this section, we present a new MILP model for solving MDMWNPP, labeled by New.
Variables introduced in the model are as follows:
• Binary variables xi j : xi j = 1, if the i-th vector is included in the j-th subset, otherwise
0;
• real variables yl , zl : maximum and minimum sum of l-th coordinate among all subsets,
more precisely:
⎧ ⎫
⎨ ⎬
yl = max vil (12)
j∈{1,...,k} ⎩ ⎭
vi ∈S j

and
⎧ ⎫
⎨ ⎬
zl = min vil (13)
j∈{1,...,k} ⎩ ⎭
vi ∈S j

• a positive real variable r : corresponds to the objective function, i.e.,

r= max {yl − zl } (14)
l∈{1,...,m}

Now, the model New is stated as follows:

min r (15)
s.t.

k
xi j = 1, ∀i ∈ {1, . . . , n} (16)
j=1

n
xi j ≥ 1, ∀ j ∈ {1, . . . , k} (17)
i=1
n
vil · xi j ≤ yl , ∀ j ∈ {1, . . . , k}, ∀l ∈ {1, . . . , m} (18)
i=1
n
vil · xi j ≥ zl , ∀ j ∈ {1, . . . , k}, ∀l ∈ {1, . . . , m} (19)
i=1
yl − zl ≤ r , ∀l ∈ {1, . . . , m} (20)
xi j ∈ {0, 1}, ∀i ∈ {1, . . . , n}, ∀ j ∈ {1, . . . , k} (21)
yl , zl ∈ R, ∀l ∈ {1, . . . , m} (22)
r ∈ [0, +∞) (23)
The task is to minimize r ∈ [0, +∞). From (20), it is evident that r is bounded from below
by maximum difference between yl and zl , which implies that the definition of the variable

123
119 Page 6 of 72 B. Nikolic et al.

r is in line with the formula (1) and expression (14). Constraints (18) and (19) correspond
to the expressions (12) and (13) and bounds yl and zl variables, l ∈ {1, . . . , m}. Constraint
(16) ensures that each vector is assigned to exactly one subset.
The condition introduced in Faria et al. [2] version of MDTWNPP, that each of the subsets
has to be non-empty is expressed by (17). Clearly, constraints (21) ensures binary nature of
xi j variables, i ∈ {1, . . . , n}, j ∈ {1, . . . , k}.
Finally, New model has nk + 2m + 1 variables (from which nk are binary variables) and
n + k + 2km + m constraints.
Remark. Note that in the MDTWNPP literature, there is lack of consistency regarding use of
the constraint of non-emptiness of subsets for k-decompositions, modeled by constraint (17).
Still, it can be shown that this constraint is redundant under the assumption that all vector
coordinates are positive real numbers as it is the case of all the known benchmark sets from
literature. This fact is proven in Appendix B.

4 Experimental results

All proposed algorithms were implemented in C++ using GCC 5.4.2. All experiments were
performed on a cluster of machines under Linux with Intel Xeon E5649 CPUs with 2.53
GHz and a memory limit of 8 GB in single-threaded mode. The maximum computation time
allowed for each run was limited to 20 min, i.e., 1200 s.
The mathematical model proposed in this paper was implemented by using Cplex solver
version 12.5.1, with the default configuration, in C++ and GCC compiler version 5.4.2. Faria’s
et al. model and Kojic’s model have been re-implemented in the form as they presented in the
original papers. We recall that the model of Kojic would need additional constraint of non-
emptiness of subsets, but it is not required due to the characteristic of the problem instances
used in the experimental evaluation (all values of all vectors include positive real numbers).
The following models are compared:
• the model of Kojić (2010), labeled by Kojic; the model is limited to instances with k = 2;
• the model by Faria et al. [2], labeled by FdSdS;
• our model presented in Sect. 3, labeled by New.
The benchmark set used in our experiments was originally generated by Kojić (2010).
All the published papers on this topic used the same starting set of instances to generate the
final benchmark sets apart of the paper by Hacibeyoglu et al. (2014). In the set of instances
proposed by Kojić (2010), there are five types of instances, denoted as t ype ∈ {a, b, c, d, e},
that is five files in which a matrix of dimension 500 × 200 has been generated. Each element
of the matrices is a float number with four decimal places that are all randomly gener-
ated. The names of the starting instances are mdtwnpp_500_20a, mdtwnpp_500_20b,
mdtwnpp_500_20c, mdtwnpp_500_20d, mdtwnpp_500_20e, respectively. As an
example, to generate a specific instance with n = 30, m = 10, k = 5, and t ype = a, we
take a starting instance mdtwnpp_500_20a and store the upper-left matrix of dimension
30 × 10 which exactly gives us 30 vectors (rows) each of dimension 10. In the same manner,
for each of the following configurations:
• the number of vectors n ∈ {30, 40, 50, 100, 200, 300, 400, 500},
• the vector’s dimension m ∈ {2, 3, 5, 10, 15, 20},
• the number of allowed subsets k ∈ {2, 3, 4, 5, 10, 20, 30},
we generate one problem instance, which gives us, in overall, 1680 problem instances.

123
New mixed-integer linear programming model... Page 7 of 72 119

Note that, until now, there have not been performed any experiments with k ≥ 5 in the
literature. Thus, this paper further expands the experimental evaluation by considering the
problem instances with larger k, k ∈ {5, 10, 20, 30}, for the first time. Moreover, perfor-
mances of the algorithms are shown for the instances with n ∈ {30, 40} for the first time to
obtain a deeper insight into limitations of the exact solving which we found not so clear
in the literature. The executable file of our implementations and the benchmark sets can be
found at https://github.com/markodjukanovic90/mtwnpp-2021. For the complete results, we
refer readers to Appendix A.
To check the statistical significance of differences of our algorithms, we generated Figs. 5a,
6, 7 and 8. The following statistical evaluation is executed. Friedman’s test was performed
simultaneously on the groups of results for all models (three models in case of k = 2, or two
otherwise). The results are grouped w.r.t. different number of (non-empty) subsets k. Given
that in all cases the test rejected the hypothesis that the results on the same group for each
model are (statistically) equal, pairwise comparisons are performed using the Nemenyi post
hoc test, see Pohlert (2014). The outcome is shown by means of so-called critical difference
plots; we have two (comparing objective values and lower bound) for each group of results.
In short, each model is positioned in the horizontal segment according to its average ranking
concerning the considered set of instances. Then, the critical difference (CD) is computed for
a significance level of 0.05 and the performance of those algorithms that have a difference
lower than CD are considered as equal–that is, no difference of statistical significance can be
detected. This is indicated in the graphics by horizontal bars joining the respective models.
In the descriptions of the obtained results, we make a small convention. By writing the
model Kojic, FdSdS, and New, we always refer to the Cplex solver which is applied on
specific model.
Tables 1, 2, 3 and 4 present summaries over the results which compare number of best
achieved solutions, average best solutions, average lower bounds and number of optimal
solutions found, all grouped w.r.t. different k.
In Table 1, the number of achieved best results for each of the three compared algorithms
is presented. These results are given in terms of a plot by Fig. 1. The following observations
can be made:

• For k = 2, the model Kojic finds the best solutions for significantly more cases than the
model FdSdS. It seems that the performance of the model New performs worse than the
other competitors.
• For k = 3, the model FdSdS achieves best results on significantly more cases than the
competitor. The same can be argued for the instances with k = 4.
• Starting with k ≥ 5, the model New delivers best results on significantly more cases than
the competitor model FdSdS. This is especially noticeable for larger k, where the model
New shows superiority (please, see k = 30 where the best results in case of the model
New are achieved on 235 cases whereas the model FdSdS achieved it on just 46 cases).

In Table 2, average objective values for each model grouped by different k are presented.
These results are displayed in terms of a plot by Fig. 2. We observe the main conclusions:

• For k = 2, the model Kojic delivers the best average solutions quality that are slightly
better than that of the model FdSdS. However, these differences are statistically irrel-
evant, see Fig. 5a. The model New here obtains significantly worse average solutions
quality than those of the two competitors.
• For k ∈ {3, 4}, the model FdSdS obtains better average solution quality than the competi-
tor model New, these differences are not statistically significant in case of the instances

123
119 Page 8 of 72 B. Nikolic et al.

with k = 4, whereas for k = 3 they appear to be significant in favor of the results of

FdSdS model; to verify this fact, see respective Figs. 5b and 6a.
• Starting with k ≥ 5, the model New delivers better results than the competitor model
FdSdS; this difference is statistically relevant for k ≥ 10, see Figs. 7a, b and 8.
In Table 3, the number of instances where optimality could be proven within the given
time limit for each model grouped by different k are presented. These results are displayed
in terms of a plot by Fig. 3. The following observations can be made:
• For k = 2, the model Kojic is able to find a proven optimal solution on 45 instances,
whereas the model FdSdS was successful in 44 cases, and the model New in 43 cases.
• For k ∈ {3, 4, 5}, the situation is similar. FdSdS and New models were able to find a
proven optimal solution for a similar number of instances.
• For k ≥ 10, the model New is able to find a proven optimal solution for significantly
larger number of instances than the competitor. The difference is especially noticeable
for k = 20, 30.
To compare the run-times on those instances where optimality could be proven, we provide
a summary by Table 4 and display it by means of plots in Fig. 4. For the sake of a fair
comparison, we take into account only those instances where all algorithms were successful
to prove optimality (see column #common optima which shows these numbers, grouped by
different k).
• For k = 2, the model Kojic needs less runtime to prove optimality than the model FdSdS.
However, the model FdSdS needs significantly less runtime than our model New.
• For k ∈ {3, 4, 5}, the model FdSdS still delivers lower run-times or at least equally good
as the competitor model New.
• For larger k ∈ {10, 20, 30}, the model New needs much less time to prove optimality
than the model FdSdS. It is especially noticeable for k = 30 where, on the 26 common
instances, the average runtime of the model New is just 3.10s in comparison to the
model FdSdS which needs 127.40s. A possible explanation for this could be based on
the difference between the numbers of constraints of the model FdSdS and the model
New. More precisely, the difference is equal to the value m(k 2 − 3k − 1). Since m ≥ 2
is fixed, to examine the sign of this expression, it is enough to consider
 √ the expression
√ 
k − 3k − 1. It can be easily evaluated that k − 3k − 1 ≤ 0, for k ∈ 3−2 13 , 3+2 13 , i.e.,
2 2

k ∈ {2, 3} since k is an integer. For other cases of k, i.e., k ≥ 4, we have k 2 − 3k − 1 > 0

which means that for these values of k the number of constraints in the model FdSdS
is larger than in the model New. Moreover, this difference gets more substantial as k
increases. As a consequence, the performance of the model FdSdS degrades quicker
than it is the case of the model New. The same conclusion can be taken why better
average solutions quality is obtained by the model New than by the model FdSdS in
case of instances with a larger value of k.

5 Conclusions

In this paper we proposed a new MILP model to solve the MDMWNPP. The experimen-
tal evaluation from the literature has been here further expanded to consider a new set of
instances, which includes the number of vectors n ∈ {30, 40} and the number of allowed
subsets k ∈ {5, 10, 20, 30}. The proposed model has been shown as mostly competitive with

123
New mixed-integer linear programming model... Page 9 of 72 119

Table 1 Number of achieved best k Kojic FdSdS New

results
2 138 125 95
3 n/a 151 119
4 n/a 137 125
5 n/a 119 137
10 n/a 88 161
20 n/a 53 206
30 n/a 46 235

Table 2 Avg. value of the k Kojic FdSdS New

objective function
2 19,614.52941 19,717.17353 20, 602.41718
3 33,883.36726 34, 714.65171
4 44,353.14992 44, 882.12398
5 52,237.99449 51, 422.91392
10 77,569.10346 74, 297.48153
20 110,263.4594 97, 814.84431
30 158,659.1417 108, 460.7812

Table 3 Number of proved k Kojic FdSdS New

optimal solutions
2 45 44 43
3 n/a 29 31
4 n/a 18 18
5 n/a 15 14
10 n/a 8 11
20 n/a 7 13
30 n/a 30 35

Table 4 Average time to reach k Kojic FdSdS New #common optima

optima
2 126.35 139.45 200.83 42
3 n/a 144.23 210.81 28
4 n/a 290.11 293.99 17
5 n/a 261.95 360.78 14
10 n/a 444.13 358.15 7
20 n/a 436.11 309.16 7
30 n/a 127.40 3.10 26

123
119 Page 10 of 72 B. Nikolic et al.

Fig. 1 Number of achieved best results

Fig. 2 Average value of the objective function

Fig. 3 Number of proved optimal solutions

123
New mixed-integer linear programming model... Page 11 of 72 119

Fig. 4 Average time to reach common optima

1 2 3 1 2

FDSDS New FDSDS

Kojic New
(a) Objective values comparison: (b) Objective values comparison:
instances with k = 2. instances with k = 3.
Fig. 5 The CD plots showing statistical difference

1 2 1 2

New New

FDSDS FDSDS

(a) Objective values comparison: (b)Objective values comparison:

instances with k = 4. instances with k = 5.
Fig. 6 The CD plots showing statistical difference

the model from literature for smaller k, and outperforms it w.r.t. number of best achieved
solutions as well as average solutions quality for larger k. This difference was statistically
significant w.r.t. the achieved average solutions quality, according to our statistical evaluation.
In future work, we could consider different mechanisms for breaking symmetries in our
model to further boost the quality of obtained solutions. In addition, developing heuristic
methods that are hybridized with MILP could be a promising research direction to consider.

123
119 Page 12 of 72 B. Nikolic et al.

1 2 1 2

New New

FDSDS FDSDS
(a) Objective values comparison: (b) Objective values comparison:
instances with k = 10. instances with k = 20.
Fig. 7 The CD plots showing statistical differences

Fig. 8 The CD plots showing 1 2

statistical differences: Objective
values comparison for instances
with k = 30

New

FDSDS

Acknowledgements This research is partially supported by Ministry for Scientific and Technological Devel-
opment, Higher Education and Information Society, Government of Republic of Srpska, B&H under the Project
“Development of artificial intelligence methods for solving computer biology problems”. The second author
was funded by the Doctoral Program Vienna Graduate School on Computational Optimization (VGSCO),
Austrian Science Foundation Project No. W1260-N35.

Appendices

Appendix A: The complete results

In this section, we provide complete results obtained by testing following MILP models:
• the model Kojic; the model is limited to the instances with k = 2;
• the model FdSdS;
• New model.
Results are organized in Tables 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21,
22, 23, 24 and 25 as follows:
• k=2
– Table 5: results for n = 30, 40, 50;
– Table 6: results for n = 100, 300, 500;
– Table 7: results for n = 200, 400.
• k=3

123
New mixed-integer linear programming model... Page 13 of 72 119

– Table 8: results for n = 30, 40, 50;

– Table 9: results for n = 100, 300, 500;
– Table 10: results for n = 200, 400.
• k=4
– Table 11: results for n = 30, 40, 50;
– Table 12: results for n = 100, 300, 500;
– Table 13: results for n = 200, 400.
• k=5
– Table 14: results for n = 30, 40, 50;
– Table 15: results for n = 100, 300, 500;
– Table 16: results for n = 200, 400.
• k = 10
– Table 17: results for n = 30, 40, 50;
– Table 18: results for n = 100, 300, 500;
– Table 19: results for n = 200, 400.
• k = 20
– Table 20: results for n = 30, 40, 50;
– Table 21: results for n = 100, 300, 500;
– Table 22: results for n = 200, 400.
• k = 30
– Table 23: results for n = 30, 40, 50;
– Table 24: results for n = 100, 300, 500;
– Table 25: results for n = 200, 400.
Each of the tables (except Tables 5, 6, 7) is organized as follows. The first three columns
represent the instance characteristics which is followed by two (or three in cases of Tables
5, 6, 7) blocks. Each block reports the results of the compared models: the model (Kojic in
case of Tables 5, 6, 7), the model Faria, and New model, respectively. Each of these blocks
consists of 3 columns which report objective value (r ), runtime (t[s]) in seconds and obtained
lower bound (LB). The minimum objective value obtained for the observed two algorithms
in Tables 8-25 (or three algorithms in case of Tables 5, 6 and 7) is highlighted in bold font,
and, if this value is of optimal quality, it is additionally underlined.

123
 Table 5 Results for n = 30, 40, 50 and k = 2
119

n m T ype Kojic FdSdS New

r t[s] LB r t[s] LB r t[s] LB

123
30 2 a 10.719 323.48 10.719 10.719 614.92 10.719 10.719 586.03 10.719
Page 14 of 72

30 2 b 17.686 369.76 17.686 17.686 436.05 17.686 17.686 548.83 17.686

30 2 c 11.857 440.07 11.857 11.857 521.74 11.857 11.857 1200 0
30 2 d 33.359 228.95 33.359 33.359 150.88 33.359 33.359 278.41 33.3581
30 2 e 12.411 151.39 12.411 12.411 161.77 12.411 12.411 444.89 12.411
30 3 a 458.689 185.61 458.689 458.689 188.48 458.689 458.689 334.05 458.689
30 3 b 374.662 120.25 374.662 374.662 157.66 374.662 374.662 277.11 374.662
30 3 c 397.202 118.19 397.202 397.202 131.4 397.202 397.202 235.24 397.202
30 3 d 431.835 200.51 431.835 431.835 227.97 431.835 431.835 438.82 431.835
30 3 e 529.719 107.92 529.719 529.719 143.79 529.719 529.719 196.8 529.719
30 5 a 2027.599 31.86 2027.599 2027.599 28.08 2027.599 2027.599 40.75 2027.599
30 5 b 4137.914 53.92 4137.914 4137.914 42.09 4137.914 4137.914 82.72 4137.914
30 5 c 5036.236 34.1 5036.236 5036.236 33.06 5036.236 5036.236 55.58 5036.236
30 5 d 5067.523 35.27 5067.523 5067.523 34.92 5067.523 5067.523 40.76 5067.523
30 5 e 4854.22 30.27 4854.22 4854.22 31.73 4854.22 4854.22 47.69 4854.22
30 10 a 35,838.597 8.68 35838.597 35,838.597 9.61 35838.597 35,838.597 29.71 35838.597
30 10 b 32,215.215 9.35 32215.215 32,215.215 6.23 32215.215 32,215.215 9.86 32215.215
30 10 c 27,339.421 6.49 27339.421 27,339.421 4.85 27339.421 27,339.421 8.62 27339.421
30 10 d 27,913.194 5.34 27913.194 27,913.194 4.05 27913.194 27,913.194 7.49 27913.194
30 10 e 31,773.234 7.33 31773.234 31,773.234 6.15 31773.234 31,773.234 13.13 31773.234
30 15 a 62,635.735 9 62635.735 62,635.735 7.68 62635.735 62,635.735 32.07 62635.735
30 15 b 58,270.95 6.92 58270.95 58,270.95 6.04 58270.95 58,270.95 9.55 58270.95
30 15 c 57,945.688 5 57945.688 57,945.688 4.22 57945.688 57,945.688 15.96 57945.688
B. Nikolic et al.
 Table 5 continued
n m T ype Kojic FdSdS New

r t[s] LB r t[s] LB r t[s] LB

30 15 d 59,496.708 6.63 59496.708 59,496.708 7.37 59496.708 59,496.708 20.34 59496.708

30 15 e 63,144.72 8.22 63144.72 63,144.72 5.86 63144.72 63,144.72 15.49 63144.72
30 20 a 78,879.971 6.43 78879.971 78,879.971 4.95 78879.971 78,879.971 20.68 78879.971
30 20 b 83,283.709 7.65 83283.709 83,283.709 11.59 83283.709 83,283.709 27.01 83283.709
30 20 c 93,979.726 16.58 93979.726 93,979.726 12.1 93979.726 93,979.726 33.69 93979.726
30 20 d 86,948.525 10.68 86948.525 86,948.525 9.54 86948.525 86,948.525 14.12 86948.525
30 20 e 83,060.654 7.19 83060.654 83,060.654 6.81 83060.654 83,060.654 10.12 83060.654
40 2 a 21.425 1200 0 27.22 1200 0 28.674 1200 0
New mixed-integer linear programming model...

40 2 b 1.525 1200 0 0.8812 1200 0 1.527 1200 0

40 2 c 1.525 1200 0 0.7003 1200 0 1.527 1200 0
40 2 d 5.739 1200 0 5.423 1200 0 8.141 1200 0
40 2 e 1.5283 1200 0 1.5214 1200 0 1.524 1200 0
40 3 a 140.937 1200 0 137.885 1200 0 137.885 1200 0
40 3 b 1.523 1200 0 3.05 1200 0 4.579 1200 0
40 3 c 1.529 1200 0 1.524 1200 0 3.048 1200 0
40 3 d 280.939 1200 0 239.539 1200 0 283.989 1200 0
40 3 e 1.5254 1200 0 4.576 1200 0 4.58 1200 0
40 5 a 3598.944 1200 0 3595.894 1200 0 3595.894 1200 0
40 5 b 3.052 1200 0 4.575 1200 0 4.576 1200 0
40 5 c 3769.863 1200 0 3769.863 1200 0 3775.969 1200 0

123
40 5 d 3204.371 1200 0 4072.443 1200 0 4072.445 1200 0
40 5 e 9.154 1200 0 3.05 1200 0 7.628 1200 0
Page 15 of 72
119
 Table 5 continued
119

n m T ype Kojic FdSdS New

r t[s] LB r t[s] LB r t[s] LB

123
40 10 a 16,516.276 345.83 16516.276 16,516.276 284.08 16516.276 16,516.276 285.48 16516.276
Page 16 of 72

40 10 b 20,877.577 501.51 20877.577 20,877.577 561.06 20877.577 20,877.577 584.42 20877.577

40 10 c 21,841.318 489.97 21841.318 21,841.318 510.28 21841.318 21,841.318 584.74 21841.318
40 10 d 17,717.383 256.48 17717.383 17,717.383 334.84 17717.383 17,717.383 758.23 17717.383
40 10 e 16,493.277 358.11 16493.277 16,493.277 323.12 16493.277 16,493.277 397.6 16493.277
40 15 a 42,266.364 279.8 42266.364 42,266.364 240.99 42266.364 42,266.364 295.24 42266.364
40 15 b 43,167.289 143.29 43167.289 43,167.289 170.24 43167.289 43,167.289 206.4 43167.289
40 15 c 44,707.051 200.82 44707.051 44,707.051 193.87 44707.051 44,707.051 204.97 44707.051
40 15 d 36,352.069 67.7 36352.069 36,352.069 57.07 36352.069 36,352.069 274.57 36352.069
40 15 e 44,955.036 262.27 44955.036 44,955.036 328.06 44955.036 44,955.036 256.71 44955.036
40 20 a 63,771.744 137.21 63771.744 63,771.744 199 63771.7432 63,771.744 663.22 63771.744
40 20 b 65,467.198 89.86 65467.198 65,467.198 104.63 65467.198 65,467.198 219.5 65467.198
40 20 c 65,383.92 153.39 65383.92 65,383.92 130.41 65383.92 65,383.92 264.35 65383.92
40 20 d 66,769.009 131.02 66769.009 66,769.009 125.19 66769.009 66,769.009 248.19 66769.009
40 20 e 63,179.416 142.63 63179.416 63,179.416 154.14 63179.416 63,179.416 257.27 63179.416
50 2 a 0.7918 1200 0 0.969 1200 0 0.973 1200 0
50 2 b 5.752 1200 0 3.309 1200 0 8.918 1200 0
50 2 c 9.189 1200 0 7.277 1200 0 9.195 1200 0
B. Nikolic et al.
 Table 5 continued
n m T ype Kojic FdSdS New

r t[s] LB r t[s] LB r t[s] LB

50 2 d 1.9507 1200 0 1.703 1200 0 7.668 1200 0

50 2 e 9.044 1200 0 9.342 1200 0 11.792 1200 0
50 3 a 291.369 1200 0 297.477 1200 0 454.112 1200 0
50 3 b 314.84 1200 0 350.834 1200 0 468.118 1200 0
50 3 c 142.937 1200 0 145.985 1200 0 145.985 1200 0
50 3 d 105.566 1200 0 105.572 1200 0 192.39 1200 0
50 3 e 220.957 1200 0 217.901 1200 0 230.107 1200 0
50 5 a 926.168 1200 0 920.066 1200 0 926.164 1200 0
New mixed-integer linear programming model...

50 5 b 3205.933 1200 0 3151.534 1200 0 3205.927 1200 0

50 5 c 2702.809 1200 0 2327.471 1200 0 2708.911 1200 0
50 5 d 2275.79 1200 0 2275.79 1200 0 2275.79 1200 0
50 5 e 4453.267 1200 0 4456.321 1200 0 4485.518 1200 0
50 10 a 16176.582 1200 0 16,173.526 1200 0 16173.53 1200 0
50 10 b 19560.314 1200 0 19,554.21 1200 0 20793.183 1200 0
50 10 c 14,134.693 1200 0 17750.997 1200 0 17750.997 1200 0
50 10 d 14921.977 1200 0 14925.025 1200 0 14,915.871 1200 0

123
Page 17 of 72
119
 Table 5 continued
119

n m T ype Kojic FdSdS New

r t[s] LB r t[s] LB r t[s] LB

123
50 10 e 19303.638 1200 0 19309.742 1200 0 15,365.957 1200 0
Page 18 of 72

50 15 a 33,208.019 1200 0 33,208.019 1200 0 33,208.019 1200 0

50 15 b 33,240.094 1200 12535.332 33,240.094 1200 6984.182 33,240.094 1200 2159.4063
50 15 c 29,456.854 1200 6112.3355 29,456.854 1200 4227.6527 29923.977 1200 0
50 15 d 21,652.841 1200 5438.0953 21,652.841 1200 5940.3798 21,652.841 1200 1998.2114
50 15 e 31,800.69 1200 0 31,800.69 1200 1559.1991 31,800.69 1200 0
50 20 a 52,826.34 1200 30251.9683 52,826.34 1200 25618.8992 52,826.34 1200 7216.5883
50 20 b 51,917.902 1200 32580.2509 51,917.902 1200 32174.3343 51,917.902 1200 15117.5657
50 20 c 50,560.864 1200 31217.3734 50,560.864 1200 31970.0527 50,560.864 1200 12409.0941
50 20 d 53,955.965 1200 30577.5703 53,955.965 1200 32981.8731 53,955.965 1200 12137.7008
50 20 e 48,281.499 1200 28009.4185 48,281.499 1200 23248.241 48,281.499 1200 13512.794
B. Nikolic et al.
 Table 6 Results for n = 100, 300, 500 and k = 2
n m t ype Kojic FdSdS New

r t[s] LB r t[s] LB r t[s] LB

100 2 a 17.303 1200 0 14.1401 1200 0 18.405 1200 0

100 2 b 3.646 1200 0 3.913 1200 0 5.539 1200 0
100 2 c 5.231 1200 0 4.53 1200 0 5.2129 1200 0
100 2 d 3.0495 1200 0 4.431 1200 0 6.433 1200 0
100 2 e 3.189 1200 0 1.071 1200 0 1.067 1200 0
100 3 a 133.309 1200 0 124.157 1200 0 298.65 1200 0
100 3 b 107.087 1200 0 97.937 1200 0 97.937 1200 0
100 3 c 266.251 1200 0 231.514 1200 0 231.512 1200 0
New mixed-integer linear programming model...

100 3 d 250.087 1200 0 247.045 1200 0 256.203 1200 0

100 3 e 150.565 1200 0 147.511 1200 0 147.513 1200 0
100 5 a 3592.841 1200 0 2453.685 1200 0 2447.591 1200 0
100 5 b 2825.846 1200 0 2077.632 1200 0 2086.786 1200 0
100 5 c 2839.322 1200 0 2788.966 1200 0 2845.426 1200 0
100 5 d 2978.991 1200 0 2982.037 1200 0 2978.991 1200 0
100 5 e 3784.8 1200 0 3778.694 1200 0 3778.692 1200 0
100 10 a 16008.039 1200 0 16011.091 1200 0 16,001.935 1200 0
100 10 b 13,600.363 1200 0 18996.495 1200 0 18999.545 1200 0
100 10 c 16575.456 1200 0 15389.616 1200 0 15,386.57 1200 0
100 10 d 15,556.06 1200 0 18282.346 1200 0 15559.11 1200 0
100 10 e 15,822.361 1200 0 16513.219 1200 0 16507.127 1200 0

123
100 15 a 32,149.343 1200 0 32,149.343 1200 0 34161.617 1200 0
100 15 b 28,717.693 1200 0 28723.797 1200 0 30794.75 1200 0
Page 19 of 72
119
 Table 6 continued
119

n m t ype Kojic FdSdS New

r t[s] LB r t[s] LB r t[s] LB

123
100 15 c 32,668.12 1200 0 32674.23 1200 0 33363.204 1200 0
Page 20 of 72

100 15 d 30,693.964 1200 0 30697.008 1200 0 30697.012 1200 0

100 15 e 30268.637 1200 0 30,256.433 1200 0 30274.747 1200 0
100 20 a 51814.656 1200 0 48,152.897 1200 0 55236.884 1200 0
100 20 b 52010.092 1200 0 46688.441 1200 0 46,685.387 1200 0
100 20 c 46,912.193 1200 0 47647.322 1200 0 51880.438 1200 0
100 20 d 43,786.661 1200 0 45923.664 1200 0 50848.357 1200 0
100 20 e 46,124.493 1200 0 48073.751 1200 0 56561.872 1200 0
300 2 a 2.909 1200 0 2.66 1200 0 1.245 1200 0
300 2 b 3.643 1200 0 3.814 1200 0 3.502 1200 0
300 2 c 0.783 1200 0 2.003 1200 0 4.473 1200 0
300 2 d 2.377 1200 0 2.577 1200 0 6.531 1200 0
300 2 e 0.7692 1200 0 6.234 1200 0 10.118 1200 0
300 3 a 190.808 1200 0 158.821 1200 0 151.388 1200 0
300 3 b 113.201 1200 0 147.881 1200 0 153.985 1200 0
300 3 c 248.295 1200 0 62.615 1200 0 212.889 1200 0
300 3 d 132.863 1200 0 134.541 1200 0 53.153 1200 0
300 3 e 203.831 1200 0 208.981 1200 0 195.258 1200 0
300 5 a 9.154 1200 0 1850.81 1200 0 1869.11 1200 0
300 5 b 4342.959 1200 0 2150.515 1200 0 2147.463 1200 0
300 5 c 921.98 1200 0 2667.158 1200 0 3187.061 1200 0
B. Nikolic et al.
 Table 6 continued
n m t ype Kojic FdSdS New

r t[s] LB r t[s] LB r t[s] LB

300 5 d 2124.567 1200 0 2155.091 1200 0 2730.907 1200 0

300 5 e 1562.015 1200 0 1574.215 1200 0 3794.385 1200 0
300 10 a 14,416.036 1201.11 0 21452.305 1200.1 0 17861.274 1200 0
300 10 b 18428.266 1203.13 0 18152.52 1200.26 0 16,255.54 1200 0
300 10 c 20508.846 1200.52 0 16,816.5 1200.15 0 22092.947 1200 0
300 10 d 19229.684 1200.59 0 17,188.357 1200.18 0 18149.577 1200 0
300 10 e 21478.476 1201.55 0 18111.889 1200.06 0 16,908.62 1200 0
300 15 a 27,488.05 1200 0 33959.196 1200 0 31629.89 1200 0
New mixed-integer linear programming model...

300 15 b 40518.826 1200 0 36,869.59 1200.83 0 42190.439 1200 0

300 15 c 41365.389 1200 0 38065.407 1200 0 36,783.579 1200 0
300 15 d 39604.987 1200 0 32,730.409 1200 0 37571.36 1200 0
300 15 e 34,148.776 1200 0 37371.088 1200 0 36829.396 1200 0
300 20 a 49,011.001 1200 0 57772.474 1200 0 60016.225 1200 0
300 20 b 53183.241 1200 0 47,929.938 1200 0 56765.672 1200 0
300 20 c 53,965.75 1200 0 61778.549 1200 0 57422.802 1200 0
300 20 d 48272.879 1200 0 47,342.561 1200 0 61173.493 1200 0
300 20 e 43,641.886 1200 0 47273.654 1200 0 60025.185 1200 0
500 2 a 5.1326 1200 0 12.49 1200 0 16.3 1200 0
500 2 b 1.969 1200 0 1.677 1200 0 1.957 1200 0
500 2 c 3.097 1200 0 4.898 1200 0 4.904 1200 0

123
500 2 d 1.673 1200 0 3.187 1200 0 16.045 1200 0
Page 21 of 72
119
 Table 6 continued
119

n m t ype Kojic FdSdS New

r t[s] LB r t[s] LB r t[s] LB

123
500 2 e 3.318 1200 0 3.6968 1200 0 6.08 1200 0
Page 22 of 72

500 3 a 10.675 1200 0 138.998 1200 0 281.282 1200 0

500 3 b 129.954 1200 0 267.37 1200 0 296.543 1200 0
500 3 c 160.334 1200 0 94.262 1200 0 79 1200 0
500 3 d 185.238 1200 0 66.696 1201.12 0 57.55 1200 0
500 3 e 4.5769 1200 0 91.379 1200 0 178.662 1200 0
500 5 a 1849.275 1200 0 1742.128 1200 0 3683.8 1200 0
500 5 b 2728.162 1200 0 2727.795 1200 0 2894.805 1200 0
500 5 c 2744.85 1200 0 1900.642 1200.44 0 1912.864 1200 0
500 5 d 13.732 1200 0 2424.008 1200 0 1982.865 1200 0
500 5 e 2833.654 1204.03 0 2667.309 1200 0 3196.62 1200.09 0
500 10 a 20076.202 1202.07 0 17,695.694 1202.7 0 20070.104 1200 0
500 10 b 18,781.808 1200.01 0 21930.449 1202.76 0 20431.433 1200 0
500 10 c 20,434.322 1200.92 0 22464.14 1200 0 21892.12 1200 0
B. Nikolic et al.
 Table 6 continued
n m t ype Kojic FdSdS New

r t[s] LB r t[s] LB r t[s] LB

500 10 d 16197.362 1200.41 0 12,894.202 1200 0 21291.872 1200 0

500 10 e 19,379.451 1200.42 0 19894.967 1200.11 0 22961.576 1200 0
500 15 a 35225.494 1200 0 34,987.747 1200 0 38486.452 1200 0
500 15 b 38371.227 1200 0 31,840.388 1200 0 34864.475 1200 0
500 15 c 41517.228 1200 0 35,025.472 1200 0 44855.907 1200 0
500 15 d 37,491.829 1200 0 39253.958 1200 0 37979.168 1200 0
500 15 e 33,823.836 1200 0 43558.088 1200 0 42331.336 1200 0
500 20 a 52,880.099 1200 0 52917.094 1200 0 60012.96 1200 0
New mixed-integer linear programming model...

500 20 b 55,750.197 1200 0 58205.477 1200 0 57775.366 1200 0

500 20 c 53,450.357 1200 0 54605.922 1200 0 63364.387 1200 0
500 20 d 56,538.543 1200 0 58465.438 1200 0 62097.2 1200 0
500 20 e 48,558.694 1200 0 54778.504 1200 0 57691.895 1200 0

123
Page 23 of 72
119
 Table 7 Results for n = 200, 400 and k = 2
119

n m T ype Kojic FdSdS New

r t[s] LB r t[s] LB r t[s] LB

123
200 2 a 16.836 1200 0 19.2545 1200 0 16.531 1200 0
Page 24 of 72

200 2 b 2.485 1200 0 0.3408 1200 0 3.343 1200 0

200 2 c 3.74 1200 0 0.85 1200 0 0.85 1200 0
200 2 d 1.207 1200 0 3.423 1200 0 6.499 1200 0
200 2 e 2.907 1200 0 9.833 1200 0 4.717 1200 0
200 3 a 181.368 1200 0 294.516 1200 0 327.024 1200 0
200 3 b 137.584 1200 0 331.04 1200 0 154.969 1200 0
200 3 c 13.734 1200 0 236.269 1200 0 245.425 1200 0
200 3 d 132.859 1200 0 139.691 1200 0 140.643 1200 0
200 3 e 203.517 1200.09 0 117.488 1200.03 0 306.625 1200 0
200 5 a 1918.864 1200 0 1918.874 1200 0 1937.18 1200 0
200 5 b 2709.857 1200 0 1882.624 1200 0 1885.68 1200 0
200 5 c 10.684 1200 0 18.317 1200 0 1855.385 1200 0
200 5 d 2384.733 1200 0 2363.363 1200 0 2127.629 1200 0
200 5 e 3798.609 1200 0 3786.407 1200 0 3591.542 1200 0
B. Nikolic et al.
 Table 7 continued
n m T ype Kojic FdSdS New

r t[s] LB r t[s] LB r t[s] LB

200 10 a 17479.552 1213.33 0 16,515.071 1200 0 16551.689 1200 0

200 10 b 17806.694 1210.43 0 20734.057 1201.4 0 17,797.538 1200 0
200 10 c 16170.862 1200 0 12,788.826 1200 0 20442.573 1200 0
200 10 d 18095.191 1200 0 18,092.139 1202.57 0 22763.495 1200 0
200 10 e 18589.563 1200.21 0 18595.665 1200 0 17,818.554 1200 0
200 15 a 29,403.65 1200 0 32633.798 1200 0 35354.676 1200 0
200 15 b 32,106.644 1200.38 0 35203.966 1200 0 38227.303 1200 0
200 15 c 30,485.324 1200 0 30500.582 1200 0 37235.134 1200 0
New mixed-integer linear programming model...

200 15 d 26,480.488 1200 0 33113.727 1200 0 37907.791 1200 0

200 15 e 28630.001 1200 0 33319.974 1200 0 28,611.703 1200 0
200 20 a 53385.485 1200 0 43,788.36 1200 0 57762.378 1200 0
200 20 b 51663.038 1200 0 49,722.198 1200 0 63713.47 1200 0
200 20 c 46379.626 1200 0 45,552.728 1200 0 51760.034 1200 0
200 20 d 48,629.33 1200 0 53277.609 1200 0 58910.412 1200 0
200 20 e 50942.559 1200 0 39,420.737 1200 0 50431.937 1200 0
400 2 a 19.158 1200 0 4.886 1200 0 6.84 1200 0
400 2 b 4.144 1200 0 0.62 1200 0 6.734 1200 0
400 2 c 1.73 1200 0 6.35 1200 0 4.904 1200 0
400 2 d 0.842 1200 0 3.846 1200 0 4.495 1200 0
400 2 e 1.548 1200 0 0.891 1200 0 8.101 1200 0

123
400 3 a 198.164 1200 0 108.548 1200.16 0 90.244 1200 0
Page 25 of 72
119
 Table 7 continued
119

n m T ype Kojic FdSdS New

r t[s] LB r t[s] LB r t[s] LB

123
400 3 b 211.739 1200 0 149.585 1200 0 175.794 1200 0
Page 26 of 72

400 3 c 97.1 1200 0 206.864 1200 0 206.876 1200 0

400 3 d 251.739 1200 0 269.846 1200 0 292.841 1200 0
400 3 e 185.529 1200 0 179.417 1200 0 149.797 1200 0
400 5 a 2472.552 1200 0 2658.672 1200 0 2646.458 1200 0
400 5 b 1491.263 1200 0 2026.252 1200 0 3078.764 1200 0
400 5 c 920.063 1200 0 3162.647 1200 0 3100.008 1200 0
400 5 d 2511.647 1200 0 2690.316 1200 0 2502.495 1200 0
400 5 e 1631.646 1200 0 1622.496 1200 0 2839.667 1200 0
400 10 a 18436.079 1200.8 0 19013.111 1200.09 0 16,725.126 1200 0
400 10 b 21190.523 1200.17 0 18702.998 1200 0 17,893.723 1200 0
400 10 c 20709.911 1203.85 0 22009.308 1200.15 0 20,391.976 1200 0
400 10 d 13735.6 1200 0 13,726.44 1200.11 0 18123.474 1200 0
B. Nikolic et al.
 Table 7 continued
n m T ype Kojic FdSdS New

r t[s] LB r t[s] LB r t[s] LB

400 10 e 22024.199 1201.54 0 18,896.512 1200 0 19541.989 1200 0

400 15 a 37709.021 1200 0 30,462.143 1200 0 38765.615 1200 0
400 15 b 38534.025 1200.26 0 34,731.633 1200.09 0 37984.444 1200 0
400 15 c 38658.121 1200 0 41839.477 1200 0 36,866.684 1200 0
400 15 d 34,646.147 1200 0 39582.518 1200 0 47523.21 1200 0
400 15 e 33711.591 1200.42 0 33,624.756 1200 0 38910.69 1200 0
400 20 a 62625.795 1200 0 62211.657 1200 0 57,268.898 1200 0
400 20 b 56002.291 1200 0 54,093.022 1200 0 55606.06 1200 0
New mixed-integer linear programming model...

400 20 c 61141.342 1200 0 59878.639 1200 0 58,489.442 1200 0

400 20 d 49,621.586 1200 0 54203.142 1200 0 55671.767 1200 0
400 20 e 54,729.968 1200 0 59583.114 1200 0 61708.882 1200 0

123
Page 27 of 72
119
119 Page 28 of 72 B. Nikolic et al.

Table 8 Results for n = 30, 40, 50 and k = 3

n m T ype FdSdS New

r t[s] LB r t[s] LB

30 2 a 80.259 292.55 80.259 80.259 950.19 80.259

30 2 b 215.741 1200 0 78 281.02 78
30 2 c 144.433 285.15 144.433 144.433 386.84 144.433
30 2 d 135.582 223.79 135.582 135.582 198.95 135.582
30 2 e 129.451 724.8 129.451 129.451 151.53 129.451
30 3 a 773.446 513.78 773.4451 773.446 56.7 773.446
30 3 b 1378.271 81.68 1378.271 1378.271 90.57 1378.271
30 3 c 1012.334 96.13 1012.334 1012.334 107.21 1012.334
30 3 d 1192.211 469.36 1192.211 1192.211 462.71 1192.211
30 3 e 1202.613 88.08 1202.613 1202.613 104.51 1202.613
30 5 a 11,072.565 49.09 11072.565 11,072.565 60.42 11072.565
30 5 b 9259.137 27.76 9259.137 9259.137 92.56 9259.137
30 5 c 8484.901 21.2 8484.901 8484.901 46.64 8484.901
30 5 d 10,773.195 257.77 10773.195 10,773.195 81.51 10773.195
30 5 e 10,826.653 49.29 10826.653 10,826.653 92.58 10826.653
30 10 a 42,823.408 28.6 42823.408 42,823.408 137.18 42823.408
30 10 b 42,828.972 24.03 42828.972 42,828.972 38.28 42828.972
30 10 c 45,214.814 32.62 45214.814 45,214.814 26.54 45214.814
30 10 d 41,095.153 36.98 41095.153 41,095.153 54.95 41095.153
30 10 e 47,539.404 68.16 47539.404 47,539.404 61.12 47539.404
30 15 a 67,164.828 76.17 67164.828 67,164.828 266.96 67164.828
30 15 b 68,745.794 85.16 68745.794 68,745.794 97.53 68745.794
30 15 c 73,912.823 48.67 73912.823 73,912.823 172.4 73912.823
30 15 d 73,713.771 57.31 73713.771 73,713.771 116.77 73713.771
30 15 e 76,456.772 65.39 76456.772 76,456.772 87.5 76456.772
30 20 a 90,909.077 111.83 90909.077 90,909.077 671.22 90909.077
30 20 b 93,941.054 206.05 93941.054 93,941.054 561.78 93941.054
30 20 c 89,914.579 83.59 89914.579 89,914.579 140.22 89914.579
30 20 d 92,540.911 137.8 92540.911 92,540.911 111.6 92540.911
30 20 e 95,571.301 309.51 95571.301 95,571.301 532.54 95571.301
40 2 a 320.608 1200 0 507.208 1200 0
40 2 b 323.36 1200 0 374.141 1200 0
40 2 c 122.389 1200 0 322.694 1200 0
40 2 d 244.968 1200 0 169.011 1200 0
40 2 e 318.291 1200 0 123.328 1200 0
40 3 a 2145.481 1200 0 2135.036 1200 0
40 3 b 1515.038 1200 0 2854.636 1200 0
40 3 c 1469.431 1200 0 1108.191 1200 0
40 3 d 2157.667 1200 0 3218.618 1200 0

123
New mixed-integer linear programming model... Page 29 of 72 119

Table 8 continued
n m T ype FdSdS New

r t[s] LB r t[s] LB

40 3 e 2157.026 1200 0 2313.412 1200 0

40 5 a 8787.489 1200 0 9318.52 1200 0
40 5 b 10,144.831 1200 0 11014.181 1200 0
40 5 c 8864.727 1200 0 9265.389 1200 0
40 5 d 9575.019 1200 0 9388.777 1200 0
40 5 e 10,905.765 1200 0 14737.84 1200 0
40 10 a 27,751.902 1200 23871.0066 36844.128 1200 14691.011
40 10 b 40558.087 1200 12227.3042 40,293.232 1200 4127.8009
40 10 c 40615.414 1200 13274.2334 30,202.157 1165.8 30202.157
40 10 d 32,276.056 1200 26157.6046 35549.589 1200 12818.2165
40 10 e 42077.713 1200 5059.9956 37,283.926 1200 23039.5066
40 15 a 59,645.354 1200 40390.477 67271.245 1200 30485.574
40 15 b 65941.57 1200 40033.5941 60,322.254 1200 44435.5893
40 15 c 55,700.432 1013.33 55700.432 55,700.432 1200 50797.3659
40 15 d 60,545.075 1200 53421.8363 63928.402 1200 45564.2238
40 15 e 65523.233 1200 41455.003 62,618.34 1200 39324.6893
40 20 a 91489.348 1200 51161.6224 91,311.519 1200 41554.8479
40 20 b 89,628.016 1200 56861.4346 98972.779 1200 42560.6136
40 20 c 89721.493 1200 57697.798 88,495.385 1200 53118.2211
40 20 d 91368.765 1200 56478.1937 83,662.159 1200 61730.033
40 20 e 91,021.352 1200 55972.5547 96655.744 1200 54319.4004
50 2 a 317.893 1200 0 309.154 1200 0
50 2 b 203.182 1200 0 129.132 1200 0
50 2 c 219.844 1200 0 262.014 1200 0
50 2 d 276.328 1200 0 264.321 1200 0
50 2 e 131.002 1200 0 244.691 1200 0
50 3 a 1405.353 1200 0 1244.386 1200 0
50 3 b 1993.619 1200 0 1697.217 1200 0
50 3 c 1522.863 1200 0 617.054 1200 0
50 3 d 1801.01 1200 0 1456.272 1200 0
50 3 e 2340.947 1200 0 1573.148 1200 0
50 5 a 10,308.708 1200 0 11238.054 1200 0
50 5 b 7840.887 1200 0 12079.316 1200 0
50 5 c 10,348.392 1200 0 11772.818 1200 0
50 5 d 9094.253 1200 0 11139.726 1200 0
50 5 e 10219.303 1200 0 9845.438 1200 0
50 10 a 28,724.766 1200 0 40640.306 1200 0
50 10 b 41,959.849 1200 0 44059.321 1200 0
50 10 c 38,073.507 1200 0 44430.482 1200 0
50 10 d 31,457.444 1200 0 42588.436 1200 0

123
119 Page 30 of 72 B. Nikolic et al.

Table 8 continued
n m T ype FdSdS New

r t[s] LB r t[s] LB

50 10 e 39,290.617 1200 0 42942.348 1200 0

50 15 a 63,736.587 1200 0 69707.444 1200 0
50 15 b 67,792.368 1200.19 0 68758.332 1200 3443.2027
50 15 c 67,964.141 1200 0 71341.735 1200 0
50 15 d 69, 620 1200 0 59,654.074 1200 0
50 15 e 59,269.802 1200 0 68855.733 1200 0
50 20 a 96373.413 1200 9540.3793 91,361.477 1200 7348.0489
50 20 b 91,380.297 1200 11181.5726 89,459.974 1200 1739.7672
50 20 c 94,780.729 1200 13746.2538 97502.087 1200 13049.1647
50 20 d 83,460.991 1200 18177.2677 87870.423 1200 12117.2923
50 20 e 85,985.393 1200 12698.6859 90422.946 1200 12051.6278

Table 9 Results for n = 100, 300, 500 and k = 3

n m T ype FdSdS New

r t[s] LB r t[s] LB

100 2 a 193.192 1200.1 0 92.688 1200 0

100 2 b 161.118 1200 0 167.797 1200 0
100 2 c 148.432 1200 0 157.28 1200 0
100 2 d 233.734 1200 0 160.461 1200 0
100 2 e 255.704 1200 0 375.311 1200 0
100 3 a 1249.459 1200 0 1208.584 1200 0
100 3 b 1697.015 1200 0 1158.115 1200 0
100 3 c 1411.766 1200 0 1550.38 1200 0
100 3 d 1529.636 1200 0 1403.444 1200 0
100 3 e 1668.658 1200 0 1428.696 1200 0
100 5 a 10,418.016 1200 0 11028.8 1200 0
100 5 b 11,399.312 1200 0 11433.442 1200.3 0
100 5 c 9443.727 1200 0 11893.044 1200 0
100 5 d 12551.612 1200 0 8808.212 1200 0
100 5 e 10884.699 1200 0 10,456.419 1200 0
100 10 a 39,730.923 1200 0 44090.095 1200 0
100 10 b 33,281.848 1200 0 38945.758 1200 0
100 10 c 40129.175 1200 0 39,453.714 1200 0
100 10 d 38,729.231 1200 0 41833.15 1200 0
100 10 e 39419.147 1200 0 38,264.1 1200 0
100 15 a 59,056.436 1200 0 60969.112 1200 0

123
New mixed-integer linear programming model... Page 31 of 72 119

Table 9 continued
n m T ype FdSdS New

r t[s] LB r t[s] LB

100 15 b 67856.899 1200 0 62,057.077 1200 0

100 15 c 59762.607 1200 0 55,986.222 1200 0
100 15 d 68501.063 1200 0 58,434.533 1200 0
100 15 e 74391.616 1200 0 59,853.418 1200 0
100 20 a 95286.274 1200 0 85,561.934 1200 0
100 20 b 83,957.155 1200 0 92309.958 1200 0
100 20 c 86,583.714 1200 0 89632.159 1200 0
100 20 d 86,749.649 1200 0 92177.919 1200 0
100 20 e 88077.106 1200 0 86,400.076 1200 0
300 2 a 202.223 1200.94 0 239.22 1205.04 0
300 2 b 287.741 1200 0 169.5 1204.07 0
300 2 c 66.088 1201.25 0 127.907 1200 0
300 2 d 305.74 1200.37 0 170.905 1204.66 0
300 2 e 67.659 1200 0 174.28 1200 0
300 3 a 1625.162 1200.12 0 1256.199 1200 0
300 3 b 1772.102 1200 0 2030.697 1200 0
300 3 c 842.076 1200 0 1565.426 1200 0
300 3 d 1980.135 1200 0 2317.517 1200 0
300 3 e 1647.167 1200.23 0 2534.023 1200 0
300 5 a 9535.332 1200 0 11754.545 1200 0
300 5 b 10410.224 1200 0 9616.228 1200 0
300 5 c 8799.906 1200 0 9642.188 1200.26 0
300 5 d 10,916.582 1200 0 14517.019 1200.07 0
300 5 e 9601.762 1200 0 9810.493 1200 0
300 10 a 36747.868 1200 0 35,056.145 1200 0
300 10 b 39436.93 1200 0 33,392.556 1200 0
300 10 c 41958.988 1200 0 29,139.115 1200 0
300 10 d 38,330.055 1200 0 42759.786 1200 0
300 10 e 40191.523 1200 0 36,208.151 1200 0
300 15 a 66082.967 1200 0 57,421.682 1200 0
300 15 b 50,990.525 1200 0 61602.229 1200 0
300 15 c 66,680.359 1200 0 68861.33 1200 0
300 15 d 68,634.577 1200 0 71338.187 1200 0
300 15 e 63,722.366 1200 0 67399.53 1200 0
300 20 a 90556.088 1200 0 90,005.626 1200 0
300 20 b 93,738.707 1200 0 101511.242 1200 0

123
119 Page 32 of 72 B. Nikolic et al.

Table 9 continued
n m T ype FdSdS New

r t[s] LB r t[s] LB

300 20 c 89663.736 1200 0 88,623.772 1200 0

300 20 d 89832.556 1200 0 88,468.376 1200 0
300 20 e 88,725.587 1200 0 94730.287 1200 0
500 2 a 123.884 1208.21 0 162.664 1200 0
500 2 b 279.341 1208.87 0 262.908 1200 0
500 2 c 160.498 1206.03 0 258.503 1200 0
500 2 d 225.848 1205.93 0 134.418 1209.13 0
500 2 e 103.044 1206.33 0 39.711 1200 0
500 3 a 1380.356 1200.64 0 1534.254 1200 0
500 3 b 1723.319 1204.74 0 1643.997 1200 0
500 3 c 1430.253 1206.72 0 885.304 1200 0
500 3 d 2025.618 1202.91 0 1531.129 1200 0
500 3 e 1415.841 1202.85 0 1604.97 1200 0
500 5 a 10880.144 1200 0 9196.418 1200 0
500 5 b 12377.821 1200 0 10,478.277 1200 0
500 5 c 7058.296 1200 0 10195.353 1200.13 0
500 5 d 9421.06 1200 0 8587.859 1200 0
500 5 e 9092.992 1200 0 10688.123 1200.05 0
500 10 a 36,935.645 1200 0 43258.487 1200 0
500 10 b 34,545.025 1200 0 43644.319 1200 0
500 10 c 36,715.003 1200 0 39497.799 1200 0
500 10 d 31,334.361 1200 0 44922.956 1200 0
500 10 e 33,245.182 1200 0 40463.256 1200 0
500 15 a 62,192.663 1200 0 73221.412 1200 0
500 15 b 62,743.504 1200 0 70867.434 1200 0
500 15 c 63,852.549 1200 0 67274.186 1200 0
500 15 d 60,902.016 1200 0 67187.431 1200 0
500 15 e 64,902.84 1200 0 65833.195 1200 0
500 20 a 85,958.535 1200.01 0 91286.375 1200 0
500 20 b 85716.808 1200 0 82,184.912 1200 0
500 20 c 71,964.643 1200 0 90785.725 1200 0
500 20 d 97886.603 1200 0 91,457.664 1200 0
500 20 e 91827.787 1200 0 91,263.67 1200 0

123
New mixed-integer linear programming model... Page 33 of 72 119

Table 10 Results for n = 200, 400 and k = 3

n m T ype FdSdS New

r t[s] LB r t[s] LB

200 2 a 159.974 1200 0 79.569 1200 0

200 2 b 136.248 1200 0 261.653 1200.43 0
200 2 c 91.293 1200 0 228.456 1200 0
200 2 d 148.054 1200.24 0 261.916 1200 0
200 2 e 180.024 1200.27 0 131.543 1200 0
200 3 a 1196.877 1200 0 631.15 1200 0
200 3 b 1125.584 1200 0 2231.055 1200 0
200 3 c 1369.679 1200 0 2466.571 1200 0
200 3 d 1687.252 1200 0 2146.71 1200 0
200 3 e 1506.17 1200 0 1545.151 1200 0
200 5 a 9764.38 1200 0 9127.63 1200 0
200 5 b 9418.789 1200 0 12411.434 1200.97 0
200 5 c 5839.467 1200 0 10195.665 1200 0
200 5 d 9873.172 1200 0 11196.528 1200.76 0
200 5 e 10,606.923 1200 0 13243.998 1201.17 0
200 10 a 36,350.164 1200 0 40544.755 1200 0
200 10 b 42913.81 1200 0 39,322.039 1200 0
200 10 c 34,076.224 1200 0 36116.264 1200 0
200 10 d 37,926.069 1200 0 42110.976 1200 0
200 10 e 45851.853 1200 0 42,212.323 1200 0
200 15 a 54,012.679 1200 0 69195.547 1200 0
200 15 b 62910.295 1200 0 62,388.548 1200 0
200 15 c 58,550.943 1200 0 65953.103 1200 0
200 15 d 65,169.349 1200 0 71979.057 1200 0
200 15 e 63,557.545 1200 0 70975.367 1200 0
200 20 a 85440.475 1200 0 80,396.46 1200 0
200 20 b 89580.693 1200 0 82,141.502 1200 0
200 20 c 74,128.064 1200 0 85432.046 1200 0
200 20 d 84,400.188 1200 0 86626.178 1200 0
200 20 e 88,915.728 1200 0 92624.3 1200 0
400 2 a 227.071 1203.63 0 89.081 1200 0
400 2 b 192.535 1203.95 0 176.891 1200 0
400 2 c 273.407 1201.1 0 81.132 1208.28 0
400 2 d 360.078 1202.35 0 263.478 1200 0

123
119 Page 34 of 72 B. Nikolic et al.

Table 10 continued
n m T ype FdSdS New

r t[s] LB r t[s] LB

400 2 e 124.102 1202.55 0 194.826 1200.13 0

400 3 a 1072.531 1200 0 1128.616 1200 0
400 3 b 1840.54 1201.92 0 2462.509 1200 0
400 3 c 1190.525 1203.14 0 2436.516 1200 0
400 3 d 794.487 1202.08 0 2131.033 1200 0
400 3 e 2208.49 1200.22 0 2240.918 1200 0
400 5 a 10065.049 1200 0 9628.135 1200.11 0
400 5 b 8620.8 1200 0 10819.646 1200.28 0
400 5 c 9930.805 1200 0 7145.684 1200.15 0
400 5 d 6414.779 1200 0 12751.181 1200 0
400 5 e 10,638.446 1200 0 12737.588 1200.17 0
400 10 a 33062.416 1200 0 29,030.453 1200 0
400 10 b 36,772.451 1200 0 44567.119 1200 0
400 10 c 42124.733 1200 0 37,127.974 1200 0
400 10 d 35,066.124 1200 0 46377.238 1200 0
400 10 e 36,408.221 1200 0 37784.038 1200 0
400 15 a 64,132.673 1200 0 64629.296 1200 0
400 15 b 66505.745 1200 0 62,238.581 1200 0
400 15 c 75579.191 1200 0 58,805.149 1200 0
400 15 d 67175.98 1200 0 60,406.197 1200 0
400 15 e 66247.82 1200 0 60,822.148 1200 0
400 20 a 81817.325 1200 0 79,522.755 1200 0
400 20 b 85,452.88 1200 0 85940.935 1200 0
400 20 c 90070.181 1200 0 87,392.614 1200 0
400 20 d 95,652.502 1200 0 96400.061 1200 0
400 20 e 78,702.317 1200 0 93572.512 1200 0

123
New mixed-integer linear programming model... Page 35 of 72 119

Table 11 Results for n = 30, 40, 50 and k = 4

n m T ype FdSdS New

r t[s] LB r t[s] LB

30 2 a 476.461 1200 381.7173 476.461 704.22 476.461

30 2 b 434.266 275.54 434.266 434.266 144.36 434.266
30 2 c 690.456 1200 214.4163 493.275 1200 438.447
30 2 d 404.159 276.14 404.159 404.159 190.53 404.159
30 2 e 486.858 253.55 486.858 486.858 360.75 486.858
30 3 a 2691.514 1028.84 2691.514 2691.514 82.41 2691.514
30 3 b 3859.603 307.41 3859.603 3859.603 352.29 3859.603
30 3 c 3312.903 116.57 3312.903 3312.903 333.56 3312.903
30 3 d 2996.785 134.05 2996.785 2996.785 71.03 2996.785
30 3 e 3164.05 163.51 3164.05 3164.05 161.84 3164.05
30 5 a 15,662.239 87.55 15662.239 15,662.239 472.48 15662.239
30 5 b 16,329.085 130.58 16329.085 16,329.085 106.47 16329.085
30 5 c 14,719.782 81.17 14719.782 14,719.782 346.55 14719.782
30 5 d 18,241.625 221.3 18241.625 18,241.625 148.38 18241.625
30 5 e 16,472.702 148.37 16472.702 16,472.702 141.69 16472.702
30 10 a 47,407.536 103.62 47407.536 47,407.536 410.12 47407.536
30 10 b 60,485.478 786.66 60485.478 60,485.478 574.22 60485.478
30 10 c 59,732.302 719.41 59732.302 59,732.302 1200 59164.7272
30 10 d 42,615.872 412.98 42615.872 42,615.872 183.89 42615.872
30 10 e 56,243.651 404.07 56243.651 56,243.651 917.19 56243.651
30 15 a 86159.422 1200 63695.4803 86,015.468 1200 64446.6463
30 15 b 89,239.516 1200 72923.299 92378.586 1200 74788.0937
30 15 c 88,457.13 1200 79756.5843 88,457.13 1200 74777.486
30 15 d 84,269.4 1200 65954.8538 84,269.4 1200 71393.8213
30 15 e 87270.621 1200 77814.3283 86,602.419 1200 77294.9317
30 20 a 106,025.717 1200 88029.6953 107436.246 1200 81656.4923
30 20 b 107,509.547 1200 87025.5497 111681.539 1200 88054.4327
30 20 c 108128.536 1200 94684.4633 107,783.722 1200 95923.0087
30 20 d 111428.33 1200 91612.321 110,715.98 1200 85141.7088
30 20 e 102,003.679 1200 91387.1624 102,003.679 1200 84640.2681

123
119 Page 36 of 72 B. Nikolic et al.

Table 11 continued
n m T ype FdSdS New

r t[s] LB r t[s] LB

40 2 a 1293.599 1200 0 1274.943 1200 0

40 2 b 897.264 1200 0 1130.51 1200 0
40 2 c 888.049 1200 0 1103.024 1200 0
40 2 d 1052.271 1200 0 1079.849 1200 0
40 2 e 915.122 1200 0 962.333 1200 0
40 3 a 5105.505 1200 0 3483.441 1200 0
40 3 b 6004.476 1200 0 7708.947 1200 0
40 3 c 4915.436 1200 0 4914.408 1200 0
40 3 d 5394.131 1200 0 6295.255 1200 0
40 3 e 4364.643 1200 0 6119.401 1200 0
40 5 a 16744.835 1200 0 13,762.232 1200 0
40 5 b 20117.902 1200 0 18,911.636 1200 0
40 5 c 18,139.451 1200 0 18363.485 1200 0
40 5 d 23,293.091 1200 0 24867.165 1200 0
40 5 e 22559.158 1200 3322.2897 22,149.327 1200 0
40 10 a 49129.67 1200 23165.883 48,258.578 1200 18737.0289
40 10 b 50748.499 1200 24309.4807 40,172.57 1200 25666.5327
40 10 c 63658.595 1200 19633.9605 55,516.159 1200 15704.9368
40 10 d 53190.053 1200 14954.3794 51,738.302 1200 22756.482
40 10 e 53,060.095 1200 24687.9005 61895.388 1200 18892.259
40 15 a 79,138.683 1200 39194.4547 84909.701 1200 33933.6584
40 15 b 77,920.308 1200 37952.2043 87096.929 1200 31323.7889
40 15 c 78,489.705 1200 40067.8451 91077.877 1200 30967.9922
40 15 d 87,568.196 1200 33599.2379 97105.447 1200 40014.9527
40 15 e 86,758.055 1200 31712.8129 87387.643 1200 36756.916
40 20 a 103,991.524 1200 40552.5475 109272.442 1200 38878.7293
40 20 b 111,111.261 1200 55677.5169 114917.443 1200 45420.6349
40 20 c 105,854.558 1200 39491.3997 118919.746 1200 44030.8456
40 20 d 103,465.009 1200 44064.186 108646.541 1200 35971.6247
40 20 e 109577.983 1200 39935.9243 106,184.045 1200 35723.7755
50 2 a 1270.716 1200 0 1190.588 1200 0

123
New mixed-integer linear programming model... Page 37 of 72 119

Table 11 continued
n m T ype FdSdS New

r t[s] LB r t[s] LB

50 2 b 1010.463 1200 0 568.648 1200 0

50 2 c 1045.69 1200 0 586.357 1200 0
50 2 d 954.772 1200 0 935.151 1200 0
50 2 e 927.791 1200 0 845.418 1200 0
50 3 a 4970.049 1200 0 5623.479 1200 0
50 3 b 5365.222 1200 0 5417.589 1200 0
50 3 c 3773.888 1200 0 4417.341 1200 0
50 3 d 4043.057 1200 0 4755.618 1200 0
50 3 e 6537.878 1200 0 3950.427 1200 0
50 5 a 20847.197 1200 0 15,534.998 1200 0
50 5 b 20360.081 1200 0 15,395.31 1200 0
50 5 c 20833.769 1200 0 19,874.039 1200 0
50 5 d 18,438.365 1200 0 24473.527 1200 0
50 5 e 25379.872 1200 0 21,132.106 1200 0
50 10 a 51730.47 1200 0 50,849.672 1200 0
50 10 b 56,732.249 1200 0 57866.583 1200 0
50 10 c 56,862.357 1200 0 58387.013 1200 0
50 10 d 55,202.779 1200 0 60693.211 1200 0
50 10 e 56228.286 1200 0 55,308.656 1200 0
50 15 a 85889.285 1200 730.2231 79,702.344 1200 8550.6077
50 15 b 87960.652 1200 3698.5228 85,350.282 1200 9278.318
50 15 c 92169.504 1200 0 85,913.35 1200 5284.692
50 15 d 95542.932 1200 551.3026 85,979.168 1200 4212.3993
50 15 e 86,218.14 1200 4160.976 88546.938 1200 0
50 20 a 101,370.594 1200 20323.0975 109674.131 1200 10018.0514
50 20 b 116428.597 1200 16171.2156 102,743.811 1200 15967.182
50 20 c 114529.921 1200 14384.8757 113,852.427 1200 17871.1492
50 20 d 113643.808 1200 13960.6758 109,217.665 1200 17147.7899
50 20 e 107,132.182 1200 7415.911 107829.074 1200 9214.4655

123
119 Page 38 of 72 B. Nikolic et al.

Table 12 Results for n = 100, 300, 500 and k = 4

n m T ype FdSdS New

r t[s] LB r t[s] LB

100 2 a 595.724 1200 0 956.496 1200 0

100 2 b 714.742 1200 0 871.927 1200 0
100 2 c 363.377 1200 0 566.449 1200 0
100 2 d 563.518 1200 0 676.398 1200 0
100 2 e 496.016 1200 0 616.444 1200 0
100 3 a 4939.118 1200 0 2658.49 1200 0
100 3 b 2479.666 1200 0 3255.663 1200 0
100 3 c 4143.471 1200 0 3272.253 1200 0
100 3 d 4629.729 1200 0 3296.405 1200 0
100 3 e 3963.063 1200 0 3473.167 1200 0
100 5 a 16172.561 1200 0 15,846.7 1200 0
100 5 b 18,907.036 1200 0 20534.701 1200 0
100 5 c 16,945.923 1200 0 17692.02 1200 0
100 5 d 20,029.396 1200.01 0 21602.706 1200 0
100 5 e 19,208.476 1200 0 19349.279 1200 0
100 10 a 54,141.023 1200 0 61419.317 1200 0
100 10 b 55077.644 1200 0 53,654.723 1200 0
100 10 c 54,477.969 1200 0 62963.804 1200 0
100 10 d 52,962.124 1200 0 57517.593 1200 0
100 10 e 55841.734 1200 0 53,968.704 1200 0
100 15 a 91919.613 1200 0 70,988.514 1200 0
100 15 b 84651.933 1200 0 83,342.016 1200 0
100 15 c 79786.998 1200 0 75,650.515 1200 0
100 15 d 90022.191 1200 0 82,535.41 1200 0
100 15 e 91503.321 1200 0 84,928.803 1200 0
100 20 a 108,824.236 1200 0 111404.082 1200 0
100 20 b 104658.659 1200 0 101,590.043 1200 0
100 20 c 107,854.678 1200 0 117774.724 1200 0
100 20 d 85,110.668 1200 0 102187.647 1200 0
100 20 e 107521.91 1200 0 103,551.823 1200 0
300 2 a 356.348 1200 0 653.277 1200.36 0
300 2 b 332.779 1200 0 770.923 1200 0
300 2 c 341.217 1200 0 538.076 1200 0
300 2 d 368.973 1200 0 536.085 1200 0
300 2 e 265.534 1200 0 926.708 1200 0
300 3 a 3061.626 1200 0 3878.235 1200 0
300 3 b 4656.793 1200 0 4967.458 1200 0
300 3 c 2987.569 1200 0 3230.507 1200 0
300 3 d 3320.671 1200 0 3687.999 1200 0
300 3 e 4867.315 1200 0 2503.874 1200 0
300 5 a 15,494.229 1200 0 18396.643 1200 0

123
New mixed-integer linear programming model... Page 39 of 72 119

Table 12 continued
n m T ype FdSdS New

r t[s] LB r t[s] LB

300 5 b 19,683.825 1200 0 20255.88 1200.06 0

300 5 c 9908.01 1200 0 19873.337 1200.23 0
300 5 d 15,948.913 1200 0 19808.573 1200 0
300 5 e 16,095.236 1200 0 19256.628 1200 0
300 10 a 61483.568 1200 0 51,246.59 1200 0
300 10 b 59882.383 1200 0 45,416.699 1200 0
300 10 c 55372.209 1200 0 50,915.561 1200 0
300 10 d 47,100.92 1200 0 52007.031 1200 0
300 10 e 40,091.803 1200 0 59154.964 1200 0
300 15 a 80227.971 1200 0 77,845.709 1200 0
300 15 b 81,761.657 1200.01 0 88537.173 1200 0
300 15 c 81,371.639 1200 0 86517.069 1200 0
300 15 d 82,988.183 1200 0 84788.623 1200 0
300 15 e 86560.061 1200 0 84,629.765 1200 0
300 20 a 113,361.298 1200 0 116904.643 1200 0
300 20 b 104006.668 1200 0 101,363.135 1200 0
300 20 c 113599.65 1200 0 112,436.843 1200 0
300 20 d 106,537.921 1200 0 113009.361 1200 0
300 20 e 101,565.382 1200 0 107075.776 1200 0
500 2 a 488.441 1200 0 187.483 1200 0
500 2 b 358.612 1200 0 487.461 1200 0
500 2 c 521.578 1200 0 313.964 1200 0
500 2 d 622.315 1200 0 518.012 1200 0
500 2 e 239.685 1200.62 0 624.592 1200 0
500 3 a 3019.177 1200 0 2776.812 1200 0
500 3 b 3977.862 1200 0 3426.192 1200 0
500 3 c 2176.776 1200.01 0 3466.192 1200 0
500 3 d 2874.47 1200 0 5003.594 1200 0
500 3 e 3252.676 1200 0 5714.109 1200.03 0
500 5 a 17083.923 1200 0 14,662.666 1200 0
500 5 b 13,628.326 1200.68 0 13821.586 1200 0
500 5 c 15,589.318 1200 0 17023.73 1200 0
500 5 d 13,501.305 1201.99 0 18681.569 1200 0
500 5 e 18,926.81 1202.64 0 20008.499 1200 0
500 10 a 49,737.444 1200.01 0 52908.221 1200 0
500 10 b 50,432.167 1200 0 56325.795 1200 0
500 10 c 49814.366 1200 0 49,229.729 1200 0
500 10 d 44,864.73 1200 0 54889.61 1200 0
500 10 e 42,803.001 1200 0 45898.261 1200 0
500 15 a 90576.717 1200 0 73,107.323 1200 0

123
119 Page 40 of 72 B. Nikolic et al.

Table 12 continued
n m T ype FdSdS New

r t[s] LB r t[s] LB

500 15 b 79804.023 1200.01 0 76,924.702 1200 0

500 15 c 80,109.987 1200 0 90077.324 1200 0
500 15 d 83008.823 1200 0 80,939.675 1200 0
500 15 e 74,257.658 1200 0 83493.111 1200.01 0
500 20 a 106488.337 1200 0 105,849.943 1200 0
500 20 b 118156.812 1200 0 110,275.4 1200 0
500 20 c 112240.9 1200 0 109,250.435 1200 0
500 20 d 107254.527 1200 0 103,271.843 1200 0
500 20 e 105,773.54 1200 0 107698.151 1200 0

Table 13 Results for n = 200, 400 and k = 4

n m T ype FdSdS New

r t[s] LB r t[s] LB

200 2 a 702.307 1200 0 564.955 1200 0

200 2 b 333.273 1200 0 667.393 1200 0
200 2 c 442.229 1200 0 479.365 1200 0
200 2 d 633.389 1200 0 449.81 1200 0
200 2 e 720.175 1200 0 422.577 1200 0
200 3 a 2929.723 1200 0 4218.361 1200 0
200 3 b 4047.762 1200 0 3521.923 1200 0
200 3 c 3949.6 1200 0 3139.607 1200 0
200 3 d 3455.488 1200 0 4075.068 1200 0
200 3 e 4722.186 1200 0 4527.365 1200 0
200 5 a 20537.627 1200 0 15,639.584 1200 0
200 5 b 19,806.458 1200 0 21358.7 1200 0
200 5 c 17962.712 1200 0 13,351.277 1200 0
200 5 d 16966.555 1200 0 16,301.54 1201.4 0
200 5 e 17809.808 1200 0 16,861.759 1200 0
200 10 a 48,206.927 1200 0 54829.421 1200 0
200 10 b 55,185.276 1200 0 55867.189 1200 0
200 10 c 53082.197 1200 0 47,886.633 1200 0
200 10 d 55158.827 1200 0 54,342.489 1200 0
200 10 e 50,136.101 1200 0 57582.023 1200 0
200 15 a 80,443.27 1200 0 85806.647 1200 0
200 15 b 85674.378 1200 0 84,465.181 1200 0
200 15 c 84609.145 1200 0 82,218.399 1200 0
200 15 d 85528.325 1200 0 85,189.152 1200 0

123
New mixed-integer linear programming model... Page 41 of 72 119

Table 13 continued
n m T ype FdSdS New

r t[s] LB r t[s] LB

200 15 e 87637.541 1200 0 87,183.752 1200 0

200 20 a 103,684.152 1200 0 107296.909 1200 0
200 20 b 120558.859 1200 0 111,812.132 1200 0
200 20 c 90,341.889 1200 0 119475.228 1200 0
200 20 d 92,801.027 1200.01 0 113174.703 1200 0
200 20 e 105,673.291 1200 0 106143.533 1200 0
400 2 a 362.782 1200 0 331.58 1200 0
400 2 b 420.701 1200 0 542.84 1200 0
400 2 c 605.155 1200 0 230.752 1200 0
400 2 d 362.039 1200.54 0 333.371 1200 0
400 2 e 397.252 1200 0 445.575 1200 0
400 3 a 3559.881 1200 0 2910.901 1200 0
400 3 b 3819.704 1200 0 3649.796 1200 0
400 3 c 4417.381 1200 0 4467.348 1200 0
400 3 d 3079.263 1200 0 3272.192 1200 0
400 3 e 3879.743 1200 0 2171.671 1200 0
400 5 a 17,654.703 1200 0 18880.054 1200 0
400 5 b 18,101.46 1200 0 19145.46 1200.01 0
400 5 c 16,014.152 1200 0 16051.007 1200 0
400 5 d 16064.241 1200 0 15,973.311 1200 0
400 5 e 15,683 1200 0 17458.567 1200.03 0
400 10 a 51921.18 1200 0 50,692.653 1200 0
400 10 b 55393.75 1200 0 52,357.794 1200 0
400 10 c 46,399.087 1200 0 53444.267 1200 0
400 10 d 55930.094 1200 0 47,044.923 1200 0
400 10 e 50937.271 1200 0 46,829.706 1200 0
400 15 a 77,242.294 1200 0 84042.066 1200 0
400 15 b 79220.65 1200 0 73,856.481 1200 0
400 15 c 75,928.349 1200 0 79743.783 1200 0
400 15 d 80,717.721 1200 0 85529.174 1200 0
400 15 e 85,113.266 1200.01 0 87233.423 1200 0
400 20 a 108518.022 1200 0 107,393.94 1200 0
400 20 b 111306.079 1200 0 109,474.008 1200 0
400 20 c 102,739.801 1200 0 103282.855 1200 0
400 20 d 111623.102 1200.01 0 107,568.668 1200 0
400 20 e 94,245.268 1200 0 106948.966 1200 0

123
119 Page 42 of 72 B. Nikolic et al.

Table 14 Results for n = 30, 40, 50 and k = 5

n m T ype FdSdS New

r t[s] LB r t[s] LB

30 2 a 891.409 144.74 891.409 1356.036 1200 709.6505

30 2 b 1351.374 186.63 1351.374 1351.374 346.56 1351.374
30 2 c 1082.819 277.46 1082.819 1082.819 280.85 1082.819
30 2 d 1289.324 636.95 1289.324 1289.324 238.51 1289.324
30 2 e 1289.501 162.94 1289.501 1289.501 779.5 1289.501
30 3 a 5736.417 201.4 5736.417 5736.417 117.58 5736.417
30 3 b 5767.678 225.76 5767.678 5767.678 116.17 5767.678
30 3 c 5471.117 259.5 5471.117 5471.117 257.99 5471.117
30 3 d 6839.582 228.54 6839.582 6839.582 347.14 6839.582
30 3 e 7991.13 374.68 7991.13 7991.13 336.15 7991.13
30 5 a 21,117.612 160.76 21117.612 21,117.612 204.85 21117.612
30 5 b 22,230.203 234.69 22230.203 22,230.203 253.05 22230.203
30 5 c 21,958.479 244.42 21958.479 21,958.479 242.08 21958.479
30 5 d 25,263.452 344.77 25263.452 25,263.452 699.71 25263.452
30 5 e 23,714.885 128.76 23714.885 23,714.885 830.77 23714.885
30 10 a 62,020.545 1200 52981.2699 62,020.545 1200 53063.2013
30 10 b 58,107.539 1200 54733.6968 60879.343 1200 53141.5527
30 10 c 69300.651 1200 60229.287 67,482.481 1200 66022.3467
30 10 d 64,486.754 1200 55484.3191 65241.808 1200 54259.7457
30 10 e 67712.734 1200 61943.8577 67,273.266 1200 64983.8755
30 15 a 88663.941 1200 68702.7311 88,587.947 1200 62723.2853
30 15 b 87893.29 1200 71534.2284 82,769.916 1200 73017.403
30 15 c 107114.488 1200 73542.4861 97,324.35 1200 70399.933
30 15 d 91,891.539 1200 66679.4439 91,891.539 1200 68107.4246
30 15 e 93,658.303 1200 75880.897 96734.729 1200 72455.0235
30 20 a 110,541.287 1200 77123.754 114675.691 1200 75377.5038
30 20 b 110986.236 1200 82000.831 107,852.462 1200 78085.732
30 20 c 114,367.656 1200 86783.5278 115843.711 1200 85817.054
30 20 d 115,617.832 1200 79876.294 120574.844 1200 75202.2852
30 20 e 107,697.217 1200 85948.2144 112999.481 1200 83966.8107
40 2 a 1955.671 1200 0 2356.018 1200 0
40 2 b 2287.848 1200 0 1485.355 1200 0
40 2 c 1689.939 1200 0 1795.291 1200 0
40 2 d 1880.818 1200 0 1737.118 1200 0

123
New mixed-integer linear programming model... Page 43 of 72 119

Table 14 continued
n m T ype FdSdS New

r t[s] LB r t[s] LB

40 2 e 2027.019 1200 0 1739.354 1200 0

40 3 a 7200.643 1200 505.9455 9729.708 1200 0
40 3 b 9522.165 1200 0 6523.842 1200 628.2685
40 3 c 7748.729 1200 0 7475.508 1200 0
40 3 d 8914.064 1200 0 9776.177 1200 0
40 3 e 8867.984 1200.01 0 9130.659 1200 0
40 5 a 13,912.173 1200 6303.6713 22708.596 1200 4694.9314
40 5 b 30523.208 1200 3948.9077 25,625.785 1200 5016.8055
40 5 c 17,449.709 1200 5204.0435 29095.026 1200 0
40 5 d 22,216.136 1200 0 31805.037 1200 0
40 5 e 32661.279 1200 0 27,668.84 1200 5222.4035
40 10 a 57380.345 1200 25468.0226 55,902.036 1200 26024.182
40 10 b 62,313.319 1200 29455.2395 66108.425 1200 25566.3596
40 10 c 65,519.813 1200 28522.981 66482.996 1200 28908.095
40 10 d 73449.313 1200 26494.2778 68,095.21 1200 21469.9841
40 10 e 73,325.253 1200 27790.8405 75734.973 1200 5554.3751
40 15 a 91,445.311 1200 28062.9228 96820.641 1200 28455.9132
40 15 b 91,194.427 1200 32333.9617 99858.651 1200 38266.7752
40 15 c 96,095.185 1200 26082.1019 100382.48 1200 34805.4563
40 15 d 102848.416 1200 38891.0564 96,056.1 1200 42437.6962
40 15 e 95,522.274 1200 28343.4529 100151.558 1200 37107.1998
40 20 a 115,871.303 1200 44861.6269 116379.918 1200 40021.4459
40 20 b 115,691.688 1200 51517.6843 117125.746 1200 47847.5167
40 20 c 125991.023 1200 40293.5617 115,537.077 1200 37160.4929
40 20 d 126007.885 1200 50191.3277 121,761.116 1200 44705.1709
40 20 e 122951.888 1200 46602.5711 117,509.04 1200 35417.7641
50 2 a 1741.22 1200 0 2972.38 1200 0
50 2 b 1867.984 1200 0 1892.319 1200 0
50 2 c 1807.117 1200 0 2054.746 1200 0
50 2 d 1714.102 1200 0 1841.962 1200 0
50 2 e 2234.212 1200 0 1519.651 1200 0
50 3 a 9668.464 1200 0 8455.544 1200 0
50 3 b 8191.982 1200 0 8124.133 1200 0
50 3 c 6689.257 1200 0 6871.106 1200 0
50 3 d 8530.814 1200 0 9728.342 1200 0
50 3 e 9654.501 1200 0 7612.808 1200 0
50 5 a 24,355.467 1200 0 26344.391 1200 0
50 5 b 28035.908 1200 0 27,627.385 1200 0
50 5 c 28558.454 1200 0 27,500.286 1200 0

123
119 Page 44 of 72 B. Nikolic et al.

Table 14 continued
n m T ype FdSdS New

r t[s] LB r t[s] LB

50 5 d 27,595.36 1200 0 28193.333 1200.01 0

50 5 e 36340.338 1200 0 34,790.68 1200 0
50 10 a 71374.498 1200 0 64,304.349 1200 0
50 10 b 60213.316 1200 0 59,746.907 1200 0
50 10 c 69,231.261 1200 0 78636.852 1200 0
50 10 d 70,599.268 1200 0 73925.863 1200 0
50 10 e 76372.484 1200 0 67,416.293 1200 0
50 15 a 94,815.653 1200 6423.1616 102185.759 1200 7903.9
50 15 b 97639.353 1200 13699.0234 88,449.312 1200 6142.5114
50 15 c 86,025.805 1200 3253.8467 97079.536 1200 8988.468
50 15 d 101579.097 1200 4991.5886 99,335.356 1200 8251.1848
50 15 e 93,925.875 1200 5259.3224 98033.635 1200 6398.3928
50 20 a 112340.52 1200 18923.9742 109,890.706 1200 14814.5283
50 20 b 114,887.287 1200 24246.3898 123875.647 1200 21874.7231
50 20 c 125318.261 1200 17929.7699 115,056.474 1200 19656.4877
50 20 d 114225.541 1200 17185.0033 106,141.884 1200 24274.9711
50 20 e 123446.106 1200 19660.9519 108,747.409 1200 17774.9569

Table 15 Results for n = 100, 300, 500 and k = 5

n m T ype FdSdS New

r t[s] LB r t[s] LB

100 2 a 1753.057 1200 0 1666.825 1200 0

100 2 b 1456.644 1200 0 991.751 1200 0
100 2 c 1952.464 1200 0 1497.297 1200 0
100 2 d 1399.64 1200 0 1417.34 1200 0
100 2 e 1912.683 1200 0 1441.08 1200 0
100 3 a 8182.275 1200 0 5078.36 1200 0
100 3 b 6590.428 1200 0 7444.946 1200 0
100 3 c 7883.189 1200 0 6232.255 1200 0
100 3 d 7289.946 1200 0 8045.481 1200 0
100 3 e 8939.621 1200 0 7486.493 1200 0
100 5 a 27285.2 1200 0 19,363.929 1200 0
100 5 b 26,200.544 1200 0 26583.683 1200 0
100 5 c 24411.48 1200 0 19,533.373 1200 0
100 5 d 29640.458 1200.1 0 19,394.201 1200 0

123
New mixed-integer linear programming model... Page 45 of 72 119

Table 15 continued
n m T ype FdSdS New

r t[s] LB r t[s] LB

100 5 e 24,048.758 1200 0 29406.79 1200 0

100 10 a 66570.337 1200 0 61,770.82 1200 0
100 10 b 68,104.66 1200 0 71703.817 1200 0
100 10 c 66521.635 1200 0 56,456.727 1200 0
100 10 d 70875.06 1200 0 52,141.445 1200 0
100 10 e 70,296.576 1200 0 71588.324 1200 0
100 15 a 88,954.685 1200 0 93744.832 1200 0
100 15 b 99048.668 1200 0 98,043.567 1200 0
100 15 c 95,427.824 1200 0 106176.679 1200 0
100 15 d 99974.989 1200 0 87,354.745 1200 0
100 15 e 89,184.477 1200 0 94679.467 1200 0
100 20 a 120,874.749 1200 0 126688.638 1200 0
100 20 b 119589.385 1200 0 106,870.861 1200 0
100 20 c 120,812.752 1200 0 127640.055 1200 0
100 20 d 119458.706 1200 0 108,736.931 1200 0
100 20 e 112,999.044 1200 0 124749.813 1200 0
300 2 a 1269.853 1200 0 1592.685 1200 0
300 2 b 920.436 1200 0 799.833 1200 0
300 2 c 1260.761 1200 0 801.478 1200 0
300 2 d 972.73 1200 0 723.738 1200 0
300 2 e 668.026 1200 0 1147.186 1200 0
300 3 a 5993.818 1200 0 6558.442 1200.35 0
300 3 b 6549.059 1200 0 6972.714 1200 0
300 3 c 6994.853 1200 0 4023.872 1200 0
300 3 d 8386.31 1200 0 6642.394 1200 0
300 3 e 3979.316 1200 0 8370.502 1200 0
300 5 a 21,954.511 1200 0 24755.728 1200 0
300 5 b 20,780.985 1200 0 24808.945 1201.32 0
300 5 c 25941.601 1200 0 21,942.158 1200.15 0
300 5 d 22967.776 1200 0 22,908.327 1200.47 0
300 5 e 24,998.661 1200 0 27958.306 1200 0
300 10 a 66,057.981 1200 0 66502.949 1200 0
300 10 b 66810.825 1200 0 63,371.024 1200 0
300 10 c 70523.065 1200 0 57,221.876 1200.87 0
300 10 d 60075.134 1200 0 59,930.957 1200 0
300 10 e 75343.148 1200 0 69,771.176 1200 0
300 15 a 99645.928 1200.01 0 92,866.517 1200 0
300 15 b 97758.181 1200 0 97,113.204 1200 0
300 15 c 107359.567 1200 0 101,402.98 1200 0
300 15 d 104598.806 1200 0 97,254 1200 0

123
119 Page 46 of 72 B. Nikolic et al.

Table 15 continued
n m T ype FdSdS New

r t[s] LB r t[s] LB

300 15 e 105197.678 1200.01 0 97,879.52 1200 0

300 20 a 127522.964 1200.01 0 117,801.041 1200 0
300 20 b 120493.758 1200 0 109,206.42 1200.01 0
300 20 c 119401.101 1200 0 107,088.288 1200 0
300 20 d 126165.176 1200 0 121,665.434 1200 0
300 20 e 119455.629 1200.01 0 109,607.603 1200 0
500 2 a 996.511 1200 0 701.835 1200 0
500 2 b 1038.216 1200 0 1160.437 1202.84 0
500 2 c 865.956 1200 0 1560.493 1200 0
500 2 d 1079.131 1200 0 712.291 1200 0
500 2 e 1002.797 1200 0 950.081 1200 0
500 3 a 7827.759 1200 0 5070.884 1200 0
500 3 b 5287.602 1200 0 7620.695 1200.06 0
500 3 c 4509.017 1200.01 0 4479.932 1200 0
500 3 d 5680.724 1200 0 6747.823 1200.01 0
500 3 e 6140.74 1201.21 0 6180.715 1200 0
500 5 a 18,698.696 1202.38 0 23440.767 1200 0
500 5 b 23342.768 1200 0 21,536.634 1200 0
500 5 c 22,615.969 1200.01 0 23747.863 1200 0
500 5 d 21314.06 1200 0 17,991.923 1200 0
500 5 e 20581.337 1200.01 0 18,378.399 1200.18 0
500 10 a 59,106.015 1200 0 70732.885 1200 0
500 10 b 56,279.095 1200 0 65, 423 1200 0
500 10 c 57,457.478 1200.01 0 66655.859 1200 0
500 10 d 60438.692 1200 0 59,499.086 1200 0
500 10 e 61,604.972 1200.01 0 61905.447 1200 0
500 15 a 100,211.007 1200 0 102815.529 1200 0
500 15 b 101335.124 1200 0 86,516.564 1200.01 0
500 15 c 102432.458 1200.01 0 92,899.735 1200 0
500 15 d 96792.18 1200.01 0 93,887.974 1200 0
500 15 e 95,801.196 1200.01 0 98505.624 1200 0
500 20 a 122,524.284 1200.01 0 126588.182 1200 0
500 20 b 130697.276 1200.01 0 117,255.441 1200 0
500 20 c 120,057.386 1200.01 0 126522.997 1200 0
500 20 d 121,454.931 1200 0 131500.091 1200 0
500 20 e 118,897.724 1200.01 0 122430.038 1200 0

123
New mixed-integer linear programming model... Page 47 of 72 119

Table 16 Results for n = 200, 400 and k = 5

n m T ype FdSdS New

r t[s] LB r t[s] LB

200 2 a 1412.202 1200 0 1278.489 1200 0

200 2 b 1353.401 1200 0 1565.967 1200 0
200 2 c 871.755 1200 0 509.206 1200 0
200 2 d 1604.183 1200 0 1398.813 1200 0
200 2 e 1242.338 1200 0 1623.048 1200 0
200 3 a 9597.509 1200 0 8569.376 1200 0
200 3 b 6904.299 1200 0 6363.146 1200 0
200 3 c 7908.2 1200 0 4513.367 1200.23 0
200 3 d 5336.164 1200 0 8411.871 1200 0
200 3 e 7085.501 1200 0 9632.596 1200 0
200 5 a 26493.978 1200 0 24,875.209 1200 0
200 5 b 27720.335 1200 0 25,953.985 1200.45 0
200 5 c 26,493.145 1200 0 26501.558 1200.68 0
200 5 d 26888.581 1200 0 26,353.255 1201.44 0
200 5 e 28371.453 1200 0 27,361.9 1200 0
200 10 a 66950.371 1200 0 59,866.964 1200 0
200 10 b 65603.989 1200 0 63,882.577 1200 0
200 10 c 64763.98 1200 0 55,570.486 1200 0
200 10 d 55088.744 1200 0 50,821.812 1200 0
200 10 e 59,897.967 1200 0 69190.046 1200 0
200 15 a 86,786.125 1200 0 95219.075 1200 0
200 15 b 88,354.845 1200 0 94344.372 1200 0
200 15 c 96107.31 1200 0 89,458.164 1200 0
200 15 d 99556.499 1200 0 97,094.749 1200 0
200 15 e 91,868.938 1200 0 94591.504 1200 0
200 20 a 115625.214 1200 0 113,681.354 1200 0
200 20 b 99,848.117 1200 0 126635.134 1200 0
200 20 c 122541.257 1200.01 0 121,661.115 1200 0
200 20 d 113443.96 1200 0 110,748.65 1200 0
200 20 e 120164.343 1200.01 0 113,706.458 1200 0
400 2 a 606.266 1200 0 1271.204 1200 0
400 2 b 1689.228 1200 0 763.6 1200 0
400 2 c 597.324 1200 0 1060.598 1201.47 0
400 2 d 661.702 1200.01 0 742.708 1200 0
400 2 e 973.819 1200 0 1233.92 1200.47 0
400 3 a 5902.945 1200.03 0 6807.487 1200 0
400 3 b 6657.956 1200 0 8797.939 1200 0
400 3 c 5461.741 1200 0 6218.346 1200 0
400 3 d 4985.92 1200 0 6634.411 1200 0
400 3 e 6335.469 1200 0 5497.894 1200 0
400 5 a 22,495.75 1200 0 22658.679 1200 0

123
119 Page 48 of 72 B. Nikolic et al.

Table 16 continued
n m T ype FdSdS New

r t[s] LB r t[s] LB

400 5 b 26735.781 1200 0 23,345.183 1200.03 0

400 5 c 23598.277 1200 0 20,608.989 1200 0
400 5 d 19,562.2 1200 0 25062.085 1200 0
400 5 e 23849.271 1200 0 20,779.797 1200 0
400 10 a 66,061.217 1200 0 68620.685 1200 0
400 10 b 66761.656 1200 0 63,933.479 1200 0
400 10 c 67048.645 1200 0 64,609.35 1200 0
400 10 d 61890.201 1200 0 57,656.105 1200 0
400 10 e 60455.085 1200.01 0 60,084.905 1200 0
400 15 a 97094.861 1200 0 90,109.304 1200 0
400 15 b 86,938.826 1200 0 89774.097 1200 0
400 15 c 90,524.857 1200.01 0 92884.902 1200 0
400 15 d 88785.928 1200.01 0 87,472.092 1200 0
400 15 e 94,193.778 1200 0 95603.89 1200 0
400 20 a 129515.411 1200.02 0 96,557.069 1200.01 0
400 20 b 116,896.827 1200 0 116984.352 1200 0
400 20 c 123595.013 1200 0 113,780.698 1200 0
400 20 d 128648.269 1200 0 116,658.25 1200 0
400 20 e 133451.752 1200 0 112,338.979 1200 0

Table 17 Results for n = 30, 40, 50 and k = 10

n m T ype FdSdS New

r t[s] LB r t[s] LB

30 2 a 16743.296 1200 8009.657 15,805.822 549.61 15805.822

30 2 b 14,338.294 121.34 14338.294 14,338.294 72.1 14338.294
30 2 c 10,114.876 217.73 10114.876 10,114.876 103.7 10114.876
30 2 d 15,460.745 357.69 15460.745 15,460.745 382.68 15460.745
30 2 e 25,472.952 1200 15812.7384 25,472.952 1200 20477.4021
30 3 a 24,862.404 895.03 24862.404 24,862.404 380.78 24862.404
30 3 b 27,031.183 313.15 27031.183 27,031.183 317.76 27031.183
30 3 c 28,024.872 1200 23662.1754 28,024.872 634.57 28024.872
30 3 d 28,713.894 302.73 28713.894 28,713.894 475.77 28713.894
30 3 e 35722.587 1200 25799.6708 34,320.271 1200 32699.374
30 5 a 52549.355 1200 36067.7907 50,024.911 1200 43720.957
30 5 b 53662.784 1200 40654.6242 51,908.359 829.52 51908.359
30 5 c 47,240.179 967.97 47240.179 47708.141 1200 36139.1661
30 5 d 52,348.676 901.23 52348.676 52,348.676 774.26 52348.676

123
New mixed-integer linear programming model... Page 49 of 72 119

Table 17 continued
n m T ype FdSdS New

r t[s] LB r t[s] LB

30 5 e 63550.074 1200 41897.6053 61,496.275 1200 38449.5057

30 10 a 79,838.68 1200 59188.7643 82943.184 1200 60565.4428
30 10 b 89847.954 1200 56073.5906 78,854.718 1164.2 78854.718
30 10 c 105431.087 1200 62350.5332 105,081.051 1200 56600.2839
30 10 d 92,142.036 1200 51361.6976 96606.441 1200 50945.5829
30 10 e 100755.765 1200 56436.5126 90,903.216 1200 57134.3367
30 15 a 113115.879 1200 63871.9971 109,074.393 1200 69466.6572
30 15 b 114660.351 1200 62007.852 104,998.947 1200 72967.2617
30 15 c 121774.618 1200 62832.9864 121,202.645 1200 64921.0754
30 15 d 117118.916 1200 63544.0438 113,690.304 1200 65340.8659
30 15 e 112,080.893 1200 72384.6616 112998.546 1200 69903.5466
30 20 a 127,085.015 1200 57748.7159 131281.205 1200 71068.9409
30 20 b 134270.101 1200.01 64326.7853 124,363.794 1200 72291.9948
30 20 c 146075.447 1200 60195.4972 139,585.583 1200 63786.5985
30 20 d 134737.779 1200 64889.0342 124,575.507 1200 73143.6478
30 20 e 131065.169 1200.03 71248.8252 123,481.164 1200 76527.927
40 2 a 7476.126 1200.02 903.5897 7650.608 1200 370.2398
40 2 b 9589.554 1200 976.401 4030.423 1200 2227.006
40 2 c 8959.665 1200.11 443.6914 4510.013 1200 1386.224
40 2 d 7661.001 1200 744.8182 7760.991 1200 0
40 2 e 7719.627 1200 34.3983 10796.679 1200 57.9289
40 3 a 24,322.689 1200 2432.065 24609.248 1200 3873.9721
40 3 b 19976.764 1200 5835.6451 12,355.128 1200 8519.676
40 3 c 15,493.807 1200 4603.274 21658.817 1200 0
40 3 d 20,544.016 1200 4428.8921 23742.051 1200 2933.317
40 3 e 25924.232 1200 0 18,158.855 1200 5183.5709
40 5 a 58935.906 1200.06 12268.8218 53,098.727 1200 12236.3663
40 5 b 51,491.482 1200 15514.4251 51657.648 1200 14880.4011
40 5 c 54856.708 1200 13943.89 47,867.277 1200 2575.8033
40 5 d 58798.244 1200 8597.6264 45,451.383 1200 11593.2226
40 5 e 61857.777 1200 6118.2802 47,697.53 1200 8395.5307
40 10 a 101779.746 1200 20515.7538 85,086.796 1200 36485.906
40 10 b 84,962.05 1200 34101.6898 86479.276 1200 41749.9974
40 10 c 101107.733 1200 45523.2973 96,829.735 1200 45638.5505
40 10 d 95703.676 1200 23896.1787 90,327.978 1200 34235.3444
40 10 e 92,638.59 1200 35885.0325 100924.456 1200 26446.3186
40 15 a 124284.666 1200 38727.3793 112,175.426 1200 47019.7874
40 15 b 127509.473 1200 36629.0669 124,270.311 1200 47492.5137
40 15 c 137420.673 1200 47893.5418 122,053.188 1200 47275.483

123
119 Page 50 of 72 B. Nikolic et al.

Table 17 continued
n m T ype FdSdS New

r t[s] LB r t[s] LB

40 15 d 125284.992 1200 29780.2086 113,256.68 1200 41592.7602

40 15 e 115,833.882 1200 37339.4732 119490.895 1200 50133.7485
40 20 a 144376.101 1200 35206.732 140,111.38 1200 48192.648
40 20 b 137863.526 1200.01 47452.5038 134,880.367 1200 59080.4256
40 20 c 143909.104 1200 36989.1516 140,516.345 1200 46726.7029
40 20 d 153065.349 1200 37368.6684 138,410.79 1200 44170.0835
40 20 e 137,668.811 1200 34869.1815 146217.841 1200 47126.7457
50 2 a 7825.187 1200 0 9213.474 1200 0
50 2 b 7094.15 1200.14 0 8904.598 1200 0
50 2 c 8376.015 1200 0 7024.686 1200 0
50 2 d 7462.355 1200 0 7613.919 1200 0
50 2 e 8215.921 1200 0 5622.727 1200 0
50 3 a 19,308.283 1200 0 21825.874 1200 0
50 3 b 26340.221 1200.04 0 25,583.338 1200 0
50 3 c 20410.117 1200 0 19,734.347 1200 0
50 3 d 22486.779 1200 0 18,537.054 1200 643.1786
50 3 e 24970.673 1200 0 21,150.799 1200 0
50 5 a 54108.859 1200 0 51,325.464 1200 0
50 5 b 53,863.404 1200 0 54875.399 1200 0
50 5 c 58865.321 1200.05 0 42,483.444 1200 4118.3953
50 5 d 54,780.905 1200 0 55634.6 1200 3662.5137
50 5 e 59986.841 1200 7163.1907 56,807.046 1200 0
50 10 a 92,018.735 1200 7849.1128 101400.02 1200 3943.6014
50 10 b 99,868.286 1200 14721.4268 110606.209 1200 14193.0378
50 10 c 104348.355 1200 2965.2275 101,447.605 1200 5491.5898
50 10 d 99,954.748 1200 316.135 102915.169 1200 3411.2022
50 10 e 104,689.79 1200 0 107174.111 1200 8674.0114
50 15 a 120,465.811 1200 12792.0485 123756.312 1200 18485.5736
50 15 b 124,076.118 1200 20793.0244 126429.281 1200 24565.1666
50 15 c 136527.715 1200 10053.5942 119,494.77 1200 31162.3267
50 15 d 125,795.835 1200 18681.8556 130240.951 1200 17708.1491
50 15 e 131022.359 1200 5297.0884 118,708.617 1200 27337.5399
50 20 a 142,767.696 1200.01 24010.3658 144054.429 1200 30417.1866
50 20 b 143,870.22 1200.01 23014.5446 154426.38 1200 38588.4554
50 20 c 162304.788 1200.01 23101.0829 151,491.628 1200 25059.2081
50 20 d 150571.166 1200.01 24658.2572 149,409.19 1200 29955.7514
50 20 e 154697.18 1200.01 17442.7523 142,353.148 1200 39975.2415

123
New mixed-integer linear programming model... Page 51 of 72 119

Table 18 Results for n = 100, 300, 500 and k = 10

n m T ype FdSdS New

r t[s] LB r t[s] LB

100 2 a 10217.795 1200 0 6483.596 1200 0

100 2 b 8456.231 1200 0 6095.066 1200 0
100 2 c 7716.136 1200 0 8450.859 1200 0
100 2 d 6308.793 1200.04 0 4982.22 1200 0
100 2 e 7043.447 1200.04 0 5551.36 1200 0
100 3 a 19,020.57 1200 0 23575.178 1200 0
100 3 b 22905.656 1200 0 21,585.425 1200 0
100 3 c 22,996.983 1200 0 27354.103 1200 0
100 3 d 23225.952 1200 0 21,204.957 1200 0
100 3 e 22516.75 1200.1 0 20,652.286 1200 0
100 5 a 46,686.552 1200 0 49439.904 1200 0
100 5 b 57461.065 1200 0 45,667.151 1200 0
100 5 c 48,919.067 1200 0 54558.075 1200 0
100 5 d 44,361.398 1200 0 54321.64 1200 0
100 5 e 58,217.601 1200 0 58613.784 1200 0
100 10 a 98002.595 1200.01 0 94,279.675 1200 0
100 10 b 99689.039 1200 0 98,038.033 1200 0
100 10 c 102,448.603 1200 0 102966.706 1200 0
100 10 d 97589.549 1200 0 90,686.963 1200 0
100 10 e 106363.288 1200 0 94,539.15 1200 0
100 15 a 136182.168 1200.01 0 122,351.041 1200 0
100 15 b 118,799.555 1200.01 0 128913.719 1200 0
100 15 c 139184.896 1200.02 0 132,615.718 1200 0
100 15 d 129047.329 1200.01 0 127,458.892 1200 0
100 15 e 139011.901 1200.01 0 126,846.425 1200 0
100 20 a 145,449.749 1200.01 0 147587.071 1200 0
100 20 b 163083.845 1200.01 0 144,902.832 1200 0
100 20 c 153,705.967 1200 0 155083.79 1200 0
100 20 d 156,217.662 1200 0 160926.581 1200 0
100 20 e 158437.855 1200.01 0 148,801.022 1200 0
300 2 a 4961.886 1200.16 0 5080.848 1200.01 0
300 2 b 6720.246 1200.12 0 5920.86 1200.33 0
300 2 c 5580.573 1200.19 0 6746.552 1200.26 0
300 2 d 6266.951 1200 0 5892.042 1200.01 0
300 2 e 6583.48 1200 0 5128.287 1200.11 0
300 3 a 19,819.72 1200 0 20364.328 1200.08 0
300 3 b 17,901.199 1200 0 23140.629 1200 0
300 3 c 15,566.036 1200 0 18568.673 1200 0
300 3 d 20014.683 1200 0 17,615.407 1200 0

123
119 Page 52 of 72 B. Nikolic et al.

Table 18 continued
n m T ype FdSdS New

r t[s] LB r t[s] LB

300 3 e 17229.803 1200 0 17,145.253 1200.01 0

300 5 a 50746.148 1200 0 45,931.483 1200.01 0
300 5 b 51272.991 1200.01 0 44,077.404 1200 0
300 5 c 51812.119 1200 0 44,251.874 1200 0
300 5 d 47816.913 1200 0 44,539.812 1200 0
300 5 e 51244.875 1200 0 40,937.038 1200 0
300 10 a 92,753.12 1200 0 103841.455 1200 0
300 10 b 101321.285 1200 0 100,564.042 1200 0
300 10 c 90,432.596 1200.01 0 98877.948 1200.01 0
300 10 d 101,693.301 1200 0 105908.481 1200 0
300 10 e 108755.769 1200.01 0 96,517.769 1200 0
300 15 a 136577.262 1200.02 0 134,521.208 1200 0
300 15 b 140,153.794 1200.04 0 143218.446 1200.01 0
300 15 c 141029.966 1200.03 0 130,198.209 1200 0
300 15 d 137507.656 1200.02 0 130,680.152 1200.01 0
300 15 e 133494.025 1200.02 0 130,663.166 1200.01 0
300 20 a 170145.411 1200.02 0 144,255.041 1200.01 0
300 20 b 154868.717 1200.02 0 149,490.952 1200.01 0
300 20 c 181708.414 1200.04 0 150,740.768 1200.01 0
300 20 d 165575.474 1200.03 0 153,206.214 1200 0
300 20 e 179410.19 1200.03 0 158,598.026 1200.01 0
500 2 a 5700.803 1202.37 0 5636.472 1200 0
500 2 b 4817.564 1200.08 0 5912.672 1200.01 0
500 2 c 3556.699 1200.74 0 4297.899 1200 0
500 2 d 3365.703 1200.01 0 2773.682 1200.16 0
500 2 e 3364.124 1200.19 0 5409.395 1200.01 0
500 3 a 13,118.375 1200.01 0 15930.6 1200 0
500 3 b 13553.494 1200.01 0 13,360.084 1200 0
500 3 c 17508.299 1200 0 16,890.376 1200 0
500 3 d 16894.249 1200.01 0 15,933.689 1200 0
500 3 e 19021.005 1200.01 0 13,070.063 1200.01 0
500 5 a 45053.551 1200.02 0 42,017.179 1200.01 0
500 5 b 43269.234 1200 0 40,218.557 1200 0
500 5 c 46626.338 1200.01 0 45,673.377 1200 0
500 5 d 43,697.288 1200.01 0 50252.031 1200 0
500 5 e 43,658.067 1200.01 0 45923.22 1200 0
500 10 a 96,092.4 1200 0 98007.211 1200 0
500 10 b 104203.902 1200.01 0 99,843.97 1200 0

123
New mixed-integer linear programming model... Page 53 of 72 119

Table 18 continued
n m T ype FdSdS New

r t[s] LB r t[s] LB

500 10 c 113905.74 1200.02 0 100,314.285 1200 0

500 10 d 103066.875 1200.01 0 96,176.947 1200 0
500 10 e 106866.732 1200 0 99,517.884 1200 0
500 15 a 112,134.423 1200.03 0 121912.28 1200 0
500 15 b 141722.814 1200.01 0 126,145.441 1200 0
500 15 c 132680.413 1200.02 0 112,433.881 1200.01 0
500 15 d 128,963.681 1200.01 0 135161.335 1200.01 0
500 15 e 146526.217 1200.01 0 127,221.26 1200.01 0
500 20 a 179791.117 1200.07 0 139,766 1200.01 0
500 20 b 173529.376 1200.05 0 150,091.091 1200.01 0
500 20 c 159180.16 1200.02 0 151,908.843 1200.01 0
500 20 d 168615.743 1200.02 0 158,222.647 1200.01 0
500 20 e 161578.316 1200.01 0 142,127.447 1200.01 0

Table 19 Results for n = 200, 400 and k = 10

n m T ype FdSdS New

r t[s] LB r t[s] LB

200 2 a 8342.642 1200 0 6572.731 1200 0

200 2 b 5471.875 1200 0 6557.579 1200.65 0
200 2 c 6337.022 1200 0 4419.093 1200 0
200 2 d 6747.101 1200.02 0 6192.754 1203.55 0
200 2 e 7978.879 1200 0 7277.181 1200 0
200 3 a 20195.653 1200 0 18,785.063 1200 0
200 3 b 20,727.1 1200 0 21172.585 1200 0
200 3 c 17838.778 1200 0 17,300.364 1200 0
200 3 d 25218.702 1200 0 22,439.172 1200.47 0
200 3 e 19647.917 1200 0 18,804.464 1200 0
200 5 a 52,051.462 1200 0 53845.513 1200 0
200 5 b 40,382.121 1200 0 47049.083 1200 0
200 5 c 49070.363 1200 0 39,738.35 1200 0
200 5 d 48247.48 1200.01 0 35,617.311 1200 0
200 5 e 49,375.07 1200 0 52554.684 1200 0
200 10 a 98850.362 1200 0 94,064.272 1200 0
200 10 b 104817.821 1200 0 98,862.078 1200 0
200 10 c 111578.125 1200.01 0 105,269.399 1200 0
200 10 d 102070.958 1200.01 0 95,202.969 1200.1 0
200 10 e 106729.052 1200.01 0 97,975.182 1200 0

123
119 Page 54 of 72 B. Nikolic et al.

Table 19 continued
n m T ype FdSdS New

r t[s] LB r t[s] LB

200 15 a 134902.226 1200.01 0 124,774.26 1200 0

200 15 b 139867.38 1200.02 0 126,795.768 1200 0
200 15 c 136,512.413 1200.03 0 139779.641 1200.01 0
200 15 d 121,978.338 1200.02 0 125946.505 1200 0
200 15 e 128092.551 1200.02 0 125,193.001 1200 0
200 20 a 136,374.85 1200.01 0 165447.16 1200.01 0
200 20 b 164710.982 1200.07 0 153,157.476 1200 0
200 20 c 168947.604 1200.02 0 151,106.981 1200.01 0
200 20 d 170014.179 1200 0 158,813.94 1200.01 0
200 20 e 163138.695 1200.03 0 150,987.338 1200.01 0
400 2 a 4972.049 1200.28 0 3213.235 1200.36 0
400 2 b 5764.149 1200 0 4850.378 1201.32 0
400 2 c 3358.247 1200.01 0 4890.896 1202.83 0
400 2 d 6387.233 1200 0 5039.965 1200.84 0
400 2 e 4954.328 1200 0 3524.055 1200 0
400 3 a 18,948.797 1200 0 22149.104 1200 0
400 3 b 14,149.297 1200 0 24129.418 1200.17 0
400 3 c 16,405.439 1200.01 0 17192.915 1200.01 0
400 3 d 19674.39 1200.01 0 19,173.98 1200.24 0
400 3 e 21467.393 1200 0 21,093.346 1200.66 0
400 5 a 44,988.607 1200 0 49252.943 1200 0
400 5 b 38,070.534 1200 0 51622.195 1200 0
400 5 c 40,891.673 1200 0 41452.624 1200.01 0
400 5 d 47950.876 1200 0 43,161.918 1200 0
400 5 e 40,209.675 1200.01 0 46062.431 1200 0
400 10 a 106,521.979 1200 0 107565.538 1200 0
400 10 b 98511.27 1200 0 87,803.613 1200.01 0
400 10 c 102267.964 1200.01 0 95,241.887 1200.01 0
400 10 d 93813.294 1200.02 0 93,176.117 1200 0
400 10 e 115606.38 1200.01 0 102,498.846 1200.01 0
400 15 a 141484.454 1200.03 0 117,154.978 1200.01 0
400 15 b 142438.636 1200.02 0 130,848.939 1200 0
400 15 c 137807.196 1200.02 0 131,572.004 1200 0
400 15 d 141351.358 1200.02 0 136,643.873 1200.01 0
400 15 e 135148.223 1200.02 0 121,458.613 1200.01 0
400 20 a 165711.384 1200.02 0 164,831.954 1200.01 0
400 20 b 165799.641 1200.04 0 147,932.315 1200.01 0
400 20 c 146,783.606 1200.07 0 148475.2 1200.01 0
400 20 d 160168.297 1200.11 0 156,165.317 1200.01 0
400 20 e 158539.795 1200.02 0 148,615.861 1200.01 0

123
New mixed-integer linear programming model... Page 55 of 72 119

Table 20 Results for n = 30, 40, 50 and k = 20

n m T ype FdSdS New

r t[s] LB r t[s] LB

30 2 a 62,110.281 201.03 62110.281 62,110.281 1196.86 62110.281

30 2 b 63,290.646 179.58 63290.646 63,290.646 19.59 63290.646
30 2 c 54,965.246 1200 53101.959 54,965.246 159.52 54965.246
30 2 d 65,185.379 128.75 65185.379 65,185.379 21.25 65185.379
30 2 e 67,072.783 74.06 67072.783 67,072.783 5.14 67072.783
30 3 a 76,464.457 1200 60446.4724 76,464.457 1200 62791.2844
30 3 b 82,012.052 870.96 82012.052 82,012.052 684.74 82012.052
30 3 c 72,673.436 988.25 72673.436 72,673.436 85.11 72673.436
30 3 d 79,765.282 1200 67571.2229 79,765.282 103.8 79765.282
30 3 e 79,943.575 610.12 79943.575 79,943.575 151.46 79943.575
30 5 a 90,730.863 1200 84536.1549 90,730.863 657.3 90730.863
30 5 b 102,068.245 1200 80883.2496 102,068.245 1200 84842.212
30 5 c 99523.402 1200 70426.6584 95,235.01 1200 88440.4772
30 5 d 108556.435 1200 72535.0018 102,185.82 1200 85948.2279
30 5 e 108,252.316 1200 78936.8855 108,252.316 292.46 108252.316
30 10 a 118037.213 1200.01 70236.3495 117,217.197 1200 74237.732
30 10 b 130884.393 1200.01 86823.0079 129,726.102 1200 93427.9017
30 10 c 130407.051 1200.01 75890.1428 125,698.173 1200 92132.63
30 10 d 127,270.932 1200.01 81872.442 128,875.922 1200 89514.934
30 10 e 124,219.432 1200 85735.0215 124,219.432 1200 91485.0145
30 15 a 133560.303 1200.01 68500.7102 133,268.037 1200 80006.7992
30 15 b 137839.065 1200 81919.8107 134,616.04 1200 96539.7417
30 15 c 138685.673 1200.01 78268.1528 132,465.705 1200 84175.2223
30 15 d 138215.808 1200.01 72978.3546 133,565.717 1200 92111.3242
30 15 e 140278.83 1200.01 90825.4867 136,005.333 1200 97067.0968
30 20 a 145778.996 1200.02 68083.6763 140,793.801 1200 77060.1359
30 20 b 147057.015 1200.01 89381.0109 144,450.92 1200 104005.7428
30 20 c 150623.014 1200.04 77131.6405 141,697.364 1200 96757.4365
30 20 d 145083.796 1200.01 72463.6894 137,670.065 1200 82125.1734
30 20 e 154006.34 1200.01 83705.0855 141,934.512 1200 91823.0705
40 2 a 35,259.236 1200 15348.3129 35,259.236 1200 18232.1719
40 2 b 27,900.963 1200 24072.3368 27,900.963 596.02 27900.963
40 2 c 36452.606 1200 17750.5882 35,901.091 1200 14377.7715
40 2 d 30019.654 1200 22399.197 29,810.145 1200 25910.4891
40 2 e 25457.357 1200 15714.959 24,902.634 519.94 24902.634
40 3 a 53,476.87 1200 30234.3031 53,476.87 1200 29242.6573
40 3 b 52373.906 1200 31021.0492 52,009.002 1200 30978.4565
40 3 c 52333.626 1200.01 29954.416 51,322.951 1200 37581.2979
40 3 d 51166.999 1200 26839.0619 50,757.037 1200 26603.811

123
119 Page 56 of 72 B. Nikolic et al.

Table 20 continued
n m T ype FdSdS New

r t[s] LB r t[s] LB

40 3 e 54,956.724 1200.01 29475.5672 54,956.724 1200 26045.0585

40 5 a 71618.785 1200.01 39177.2568 70,895.778 1200 45810.6773
40 5 b 77126.647 1200.01 36290.3379 72,079.643 1200 55079.7355
40 5 c 82364.14 1200 39897.0494 75,750.252 1200 49673.5283
40 5 d 80585.323 1200 45973.438 78,405.812 1200 49914.3276
40 5 e 86765.54 1200.01 36115.8354 74,442.559 1200 44640.0847
40 10 a 110928.686 1200.01 41941.462 98,692.56 1200 64431.7372
40 10 b 117338.066 1200.01 60551.9961 99,276.092 1200 65920.6089
40 10 c 122458.033 1200.01 47822.8832 113,993.7 1200 48491.3104
40 10 d 119683.455 1200.01 53286.7278 108,571.892 1200 60617.07
40 10 e 125856.314 1200.01 59393.729 114,981.315 1200 63755.946
40 15 a 131773.905 1200.01 46041.9617 118,616.149 1200 68063.7176
40 15 b 134219.252 1200.02 56047.19 115,073.484 1200 64621.7112
40 15 c 140265.136 1200.01 44557.838 129,096.208 1200 66835.0345
40 15 d 146380.564 1200.02 41215.3877 122,658.985 1200 76030.2443
40 15 e 140628.768 1200.02 59837.4527 128,210.022 1200 70540.3421
40 20 a 145588.954 1200.03 45381.1279 129,228.716 1200 63152.74
40 20 b 150451.509 1200.03 55255.6183 132,814.682 1200 73736.288
40 20 c 151663.468 1200 46488.6247 137,332.117 1200 70124.4045
40 20 d 149824.028 1200.03 46638.7442 137,847.661 1200 59878.5047
40 20 e 145202.595 1200.01 56629.6143 128,210.022 1200 71185.3062
50 2 a 19887.782 1200 0 16,728.958 1200 0
50 2 b 21,976.891 1200 0 24097.44 1200 876.1139
50 2 c 26225.25 1200.02 0 25,993.239 1200 0
50 2 d 22,257.064 1200.1 0 22,257.064 1200 837.7672
50 2 e 33584.421 1200 0 27,363.463 1200 0
50 3 a 44239.826 1200 32.081 39,665.196 1200 7825.4595
50 3 b 44754.343 1200 0 44,449.317 1200 5317.0184
50 3 c 43398.314 1200 0 37,625.522 1200 4135.379
50 3 d 42897.824 1200 0 41,395.325 1200 185.0204
50 3 e 44,240.373 1200 0 48556.236 1200 364.334
50 5 a 78475.738 1200.02 21456.4425 72,237.17 1200 24329.8635
50 5 b 82478.916 1200 20111.8367 74,205.39 1200 22746.5568
50 5 c 82264.173 1200.01 18979.6141 70,436.881 1200 20096.1113
50 5 d 84582.134 1200.01 19152.6696 82,797.698 1200 20997.2415
50 5 e 86142.625 1200 28326.1902 80,309.402 1200 25811.8336
50 10 a 120386.351 1200.01 33784.0164 107,742.195 1200 40429.4105
50 10 b 131922.899 1200.01 36196.9346 115,328.168 1200 38302.5298
50 10 c 127505.107 1200.01 29633.4083 117,584.639 1200 38751.018
50 10 d 129510.776 1200.01 30826.7637 124,205.25 1200 32018.3999

123
New mixed-integer linear programming model... Page 57 of 72 119

Table 20 continued
n m T ype FdSdS New

r t[s] LB r t[s] LB

50 10 e 130852.066 1200.01 29572.5083 117,219.18 1200 47620.4586

50 15 a 154255.682 1200.01 17702.3009 140,880.108 1200 50057.183
50 15 b 168875.902 1200.02 35317.4167 138,942.574 1200 58458.5485
50 15 c 151550.059 1200.02 34086.4567 139,857.909 1200 52033.5577
50 15 d 152174.91 1200.01 37794.7988 141,824.138 1200 55206.4462
50 15 e 162258.21 1200.02 33042.2542 137,799.942 1200 50507.2102
50 20 a 167512.789 1200.04 35114.2741 152,283.617 1200.01 41323.0018
50 20 b 173047.023 1200.04 24949.1913 159,637.004 1200.01 49441.0666
50 20 c 182541.362 1200.04 27195.4954 166,030.514 1200 49916.7115
50 20 d 167946.7 1200.04 29512.4493 156,495.701 1200 42254.0024
50 20 e 171997.321 1200.02 40596.9162 168,360.853 1200 52566.3262

Table 21 Results for n = 100, 300, 500 and k = 20

n m T ype FdSdS New

r t[s] LB r t[s] LB

100 2 a 18738.429 1200 0 13,968.951 1200 0

100 2 b 22482.807 1200 0 16,529.734 1200 0
100 2 c 20931.487 1200 0 16,841.451 1200.01 0
100 2 d 22423.517 1200 0 11,349.463 1200 0
100 2 e 18936.176 1200 0 12,142.291 1200 0
100 3 a 38,539.352 1200.01 0 42290.631 1200 0
100 3 b 48966.463 1200.16 0 38,552.843 1200 0
100 3 c 37061.202 1200 0 35,463.502 1200 0
100 3 d 38293.773 1200.06 0 31,769.314 1200 0
100 3 e 44034.531 1200.01 0 38,022.886 1200 0
100 5 a 85175.801 1200.01 0 76,251.315 1200 0
100 5 b 73562.488 1200 0 73,066.94 1200 0
100 5 c 74433.624 1200 0 66,365.87 1200 0
100 5 d 80762.599 1200.01 0 74,850.3 1200 0
100 5 e 82132.841 1200.01 0 75,694.989 1200 0
100 10 a 140150.015 1200.02 0 128,020.62 1200.01 0
100 10 b 125,837.246 1200.02 0 131197.641 1200 0
100 10 c 143351.956 1200.04 0 126,305.426 1200 0
100 10 d 128809.194 1200.02 0 114,156.828 1200 0
100 10 e 123,222.608 1200.04 0 128603.237 1200.01 0
100 15 a 161438.792 1200.04 0 157,723.852 1200 0
100 15 b 171172.854 1200.05 0 148,072.478 1200.01 0

123
119 Page 58 of 72 B. Nikolic et al.

Table 21 continued
n m T ype FdSdS New

r t[s] LB r t[s] LB

100 15 c 182801.335 1200.05 0 158,536.992 1200 0

100 15 d 175926.889 1200.06 0 152,198.384 1200.01 0
100 15 e 167921.388 1200.06 0 146,134.198 1200 0
100 20 a 206616.65 1200.05 0 174,036.686 1200.01 0
100 20 b 188,682.409 1200.06 0 190261.027 1200 17473.5647
100 20 c 186804.281 1200 0 184,190.876 1200 318.1066
100 20 d 182366.634 1200.02 0 179,675.345 1200.01 0
100 20 e 193672.78 1200.05 0 178,553.107 1200 0
300 2 a 13587.961 1200 0 12,683.912 1200.67 0
300 2 b 12067.991 1200 0 11,413.931 1200.2 0
300 2 c 12374.791 1200.01 0 12,165.247 1200.55 0
300 2 d 12,669.965 1200.01 0 14328.549 1200.01 0
300 2 e 14511.679 1200 0 11,948.574 1200.23 0
300 3 a 43371.486 1200.01 0 42,708.605 1200.01 0
300 3 b 33068.883 1200.01 0 31,405.245 1200 0
300 3 c 27,891.642 1200 0 30118.851 1200.19 0
300 3 d 28,995.472 1200.01 0 29447.739 1200.01 0
300 3 e 43838.736 1200.01 0 24,770.976 1200.01 0
300 5 a 76799.488 1200.01 0 58,054.826 1200 0
300 5 b 77768.979 1200.03 0 64,432.516 1200.01 0
300 5 c 70455.526 1200.03 0 55,092.772 1200 0
300 5 d 83507.45 1200 0 67,979.615 1200 0
300 5 e 88925.771 1200.01 0 71,910.067 1200 0
300 10 a 149973.728 1200.05 0 129,525.751 1200.01 0
300 10 b 140379.075 1200.03 0 128,580.775 1200.01 0
300 10 c 143263.489 1200.04 0 127,390.549 1200.01 0
300 10 d 135928.482 1200.01 0 111,383.207 1200.01 0
300 10 e 137790.4 1200.02 0 118,232.513 1200.01 0
300 15 a 193113.661 1200 0 158,698.701 1200.02 0
300 15 b 170506.182 1200.06 0 159,429.053 1200.02 0
300 15 c 172581.584 1200.01 0 158,557.11 1200.01 0
300 15 d 158292.211 1200.15 0 146,995.382 1200 0
300 15 e 192356.707 1200.01 0 158,134.215 1200.01 0
300 20 a 268211.104 1200.13 0 198,913.805 1200 0
300 20 b 219532.397 1200.72 0 198,926.558 1200 0
300 20 c 271096.716 1200.11 0 212,901.36 1200 0
300 20 d 251449.518 1200.23 0 209,085.073 1200.01 0
300 20 e 294829.803 1200.14 0 201,683.49 1200.01 0
500 2 a 9954.374 1200.01 0 10059.817 1200.18 0

123
New mixed-integer linear programming model... Page 59 of 72 119

Table 21 continued
n m T ype FdSdS New

r t[s] LB r t[s] LB

500 2 b 12221.512 1200.01 0 10,027.877 1200.07 0

500 2 c 9295.492 1200.01 0 8658.34 1200.23 0
500 2 d 7356.759 1200.01 0 7599.385 1200.1 0
500 2 e 12801.393 1200.03 0 9212.484 1200.01 0
500 3 a 31792.661 1200.01 0 22,172.324 1200.02 0
500 3 b 29,255.438 1200.02 0 31875.831 1200.02 0
500 3 c 24,283.547 1200.01 0 32296.877 1200.01 0
500 3 d 38227.767 1200.01 0 31,297.855 1200.02 0
500 3 e 34892.869 1200.01 0 28,231.101 1200.02 0
500 5 a 59,371.544 1200.01 0 60015.311 1200.01 0
500 5 b 72332.802 1200.01 0 56,845.908 1200.01 0
500 5 c 83072.692 1200 0 59,507.246 1200 0
500 5 d 79184.811 1200.12 0 62,866.185 1200.01 0
500 5 e 79862.741 1200.04 0 61,570.585 1200.01 0
500 10 a 149033.44 1200 0 124,370.817 1200.01 0
500 10 b 124501.653 1200.03 0 124,378.157 1200.01 0
500 10 c 147562.537 1200.03 0 116,329.42 1200.01 0
500 10 d 118326.033 1200.02 0 118,299.614 1200.01 0
500 10 e 136702.733 1200.07 0 123,108.452 1200.01 0
500 15 a 190791.348 1200.03 0 157,829.031 1200.01 0
500 15 b 165,006.578 1200.11 0 178485.957 1200.02 0
500 15 c 172750.811 1200.03 0 142,510.496 1200.01 0
500 15 d 221455.113 1200.15 0 174,144.6 1200.01 0
500 15 e 182074.971 1200.06 0 157,913.943 1200.01 0
500 20 a 287740.487 1200.03 0 184,309.406 1200.02 0
500 20 b 290779.134 1200.1 0 225,194.695 1200.01 0
500 20 c 328909.753 1200.11 0 211,188.935 1200.01 0
500 20 d 255382.67 1200.23 0 194,686.481 1200.01 0
500 20 e 328848.552 1200.34 0 216,699.094 1200.01 0

123
119 Page 60 of 72 B. Nikolic et al.

Table 22 Results for n = 200, 400 and k = 20

n m T ype FdSdS New

r t[s] LB r t[s] LB

200 2 a 19826.397 1200 0 18,610.559 1200 0

200 2 b 18001.888 1200 0 14,811.683 1200.3 0
200 2 c 16,196.436 1200.01 0 16303.185 1200 0
200 2 d 13589.955 1200 0 10,706.078 1200.44 0
200 2 e 14,274.714 1200 0 18363.904 1200.53 0
200 3 a 40273.383 1200.01 0 37,468.928 1200 0
200 3 b 41403.101 1200.01 0 41,354.665 1200 0
200 3 c 36121.35 1200.01 0 33,962.429 1200.01 0
200 3 d 34,015.349 1200 0 35403.953 1200 0
200 3 e 44769.353 1200 0 39,032.858 1200.01 0
200 5 a 75217.684 1200.02 0 72,876.448 1200 0
200 5 b 79,134.478 1200 0 81203.054 1200 0
200 5 c 74603.333 1200.01 0 72,993.909 1200 0
200 5 d 73924.891 1200.03 0 63,366.867 1200 0
200 5 e 85564.382 1200.01 0 81,900.136 1200 0
200 10 a 147948.77 1200 0 140,788.058 1200.01 0
200 10 b 133269.354 1200.04 0 131,295.814 1200.01 0
200 10 c 129398.19 1200.06 0 128,739.005 1200.01 0
200 10 d 120410.04 1200.05 0 114,383.895 1200.01 0
200 10 e 146231.893 1200.02 0 132,516.882 1200.01 0
200 15 a 180522.923 1200.06 0 167,277.556 1200.01 0
200 15 b 159,209.998 1200.05 0 171196.175 1200.02 0
200 15 c 175244.823 1200.04 0 155,955.988 1200.01 0
200 15 d 185230.055 1200.45 0 161,738.768 1200.01 0
200 15 e 177, 103 1200.06 0 162,750.618 1200.01 0
200 20 a 193,519.545 1200.07 0 194538.074 1200.01 0
200 20 b 298022.152 1200.16 0 181,157.079 1200.02 0
200 20 c 237566.267 1200.17 0 157,664.823 1200.02 0
200 20 d 228537.007 1200.1 0 177,485.602 1200.01 0
200 20 e 179,759.4 1200.11 0 189278.635 1200.02 0
400 2 a 15,412.153 1200.01 0 16650.924 1202.17 0
400 2 b 12,065.167 1200.01 0 15241.48 1200.01 0
400 2 c 10635.921 1200.01 0 8200.965 1200.01 0
400 2 d 14279.323 1200.01 0 9508.295 1200.68 0

123
New mixed-integer linear programming model... Page 61 of 72 119

Table 22 continued
n m T ype FdSdS New

r t[s] LB r t[s] LB

400 2 e 16515.233 1200.01 0 9409.827 1205.63 0

400 3 a 28,018.98 1200.01 0 35629.978 1200.01 0
400 3 b 35,481.258 1200.02 0 36812.317 1201.67 0
400 3 c 27,320.715 1200.02 0 30513.281 1200.01 0
400 3 d 36744.359 1200.01 0 32,737.294 1200.28 0
400 3 e 40335.8 1200.01 0 34,377.87 1200.02 0
400 5 a 80228.475 1200.01 0 70,385.478 1200 0
400 5 b 61953.995 1200 0 60,124.146 1200.01 0
400 5 c 72,586.122 1200.01 0 80517.669 1200.01 0
400 5 d 84600.553 1200.04 0 67,573.793 1200.01 0
400 5 e 79576.122 1200.01 0 70,139.172 1200.01 0
400 10 a 124,572.868 1200.02 0 141913.366 1200.02 0
400 10 b 125,295.709 1200.11 0 134043.391 1200.01 0
400 10 c 126,180.106 1200.03 0 139132.533 1200 0
400 10 d 131704.112 1200.05 0 115,569.289 1200.01 0
400 10 e 122094.124 1200.02 0 118,084.099 1200.01 0
400 15 a 181326.871 1200.04 0 176,194.496 1200.02 0
400 15 b 163,829.189 1200.07 0 180396.059 1200.01 0
400 15 c 155,286.165 1200.01 0 167093.368 1200.01 0
400 15 d 169874.918 1200.19 0 156,488.151 1200.01 0
400 15 e 180057.054 1200.06 0 158,648.328 1200.02 0
400 20 a 289792.015 1200.34 0 185,344.989 1200 0
400 20 b 303587.845 1200.02 0 197,993.438 1200.02 0
400 20 c 237757.881 1200.14 0 204,500.671 1200.02 0
400 20 d 313170.396 1200.13 0 208,856.222 1200.02 0
400 20 e 281467.302 1200.22 0 189,999.644 1200.01 0

123
119 Page 62 of 72 B. Nikolic et al.

Table 23 Results for n = 30, 40, 50 and k = 30

n m T ype FdSdS New

r t[s] LB r t[s] LB

30 2 a 94,160.9 7.16 94160.9 94,160.9 0.2 94160.9

30 2 b 94,629.75 7.31 94629.75 94,629.75 0.37 94629.75
30 2 c 97,381.886 7.42 97381.886 97,381.886 0.78 97381.886
30 2 d 96,557.486 5.67 96557.486 96,557.486 0.29 96557.486
30 2 e 99,440.688 23.44 99440.688 99,440.688 0.51 99440.688
30 3 a 94,160.9 11.5 94160.9 94,160.9 0.47 94160.9
30 3 b 94,629.75 14.27 94629.75 94,629.75 0.51 94629.75
30 3 c 97,381.886 12.88 97381.886 97,381.886 0.29 97381.886
30 3 d 96,557.486 11.23 96557.486 96,557.486 0.6 96557.486
30 3 e 99,440.688 13.67 99440.688 99,440.688 0.55 99440.688
30 5 a 95,895.434 48.78 95895.434 95,895.434 1.21 95895.434
30 5 b 97,263.902 43.17 97263.902 97,263.902 1.9 96290.031
30 5 c 97,381.886 29.69 97381.886 97,381.886 1.06 97381.886
30 5 d 98,658.481 20.65 98658.481 98,658.481 0.57 98658.481
30 5 e 99,440.688 23.6 99440.688 99,440.688 1.04 99440.688
30 10 a 95,973.224 60.95 95973.224 95,973.224 2.47 95973.224
30 10 b 97,396.324 65.25 97396.324 97,396.324 4.39 97396.324
30 10 c 98,602.841 187.8 98602.841 98,602.841 2.54 98602.841
30 10 d 98,658.481 115.9 98658.481 98,658.481 3.94 98658.481
30 10 e 99,440.688 129.89 99440.688 99,440.688 3.86 99440.688
30 15 a 96,736.519 165.86 96736.519 96,736.519 4.67 96736.519
30 15 b 98,089.894 223.03 98089.894 98,089.894 5.27 98089.894
30 15 c 98,602.841 195.27 98602.841 98,602.841 2.49 98350.451
30 15 d 98,658.481 427.49 98658.481 98,658.481 3.55 98658.481
30 15 e 99,440.688 272.71 99440.688 99,440.688 10.69 99440.688
30 20 a 98,548.27 794.49 98548.27 98,548.27 9.22 96736.519
30 20 b 98,462.28 463.2 98462.28 98,462.28 11.52 98462.28
30 20 c 98,602.841 541.18 98602.841 98,602.841 15.07 98350.451
30 20 d 98,658.481 628.8 98658.481 98,658.481 5.04 98658.481
30 20 e 99,440.688 334.2 99440.688 99,440.688 14.14 99440.688
40 2 a 59,599.299 1200 55197.3368 59,599.299 59.22 59599.299
40 2 b 70,356.887 1200 61407.2553 70,356.887 43.72 70356.887
40 2 c 69,257.532 1200.01 62640.2375 69,257.532 922.77 69257.532
40 2 d 71,682.73 1200.01 63359.1667 71,682.73 35.18 71682.73
40 2 e 75,388.455 1200 66889.046 75,388.455 107.86 75388.455
40 3 a 78,396.442 1200.01 72988.3663 78,396.442 108.72 78396.442
40 3 b 82,199.457 1200 66725.0827 82,199.457 1010.61 82199.457
40 3 c 78,347.277 1200.01 58850.5191 78,347.277 1200 68087.6653
40 3 d 79,856.257 1200 71153.1492 79,856.257 117.74 79856.257

123
New mixed-integer linear programming model... Page 63 of 72 119

Table 23 continued
n m T ype FdSdS New

r t[s] LB r t[s] LB

40 3 e 88590.194 1200 80708.3671 88,089.2 1200 87729.0936

40 5 a 105832.033 1200.01 51128.7963 102,479.974 1200 70465.1475
40 5 b 106745.774 1200.01 79038.4157 105,617.486 1200 80639.2285
40 5 c 93326.94 1200.01 51817.6328 90,350.566 1200 86071.9348
40 5 d 106396.002 1200.01 69656.0088 98,292.727 1200 76619.8475
40 5 e 111673.586 1200.01 72781.0343 109,073.526 1200 84173.5721
40 10 a 126595.114 1200 51157.6815 118,822.794 1200 66251.7604
40 10 b 134584.636 1200.03 77051.6363 131,606.339 1200 89295.2131
40 10 c 127608.067 1200.02 74523.4833 122,835.311 1200 76005.673
40 10 d 139358.568 1200.02 55271.821 127,055.296 1200 88268.461
40 10 e 126,976.189 1200 58273.7492 126,976.189 1200 80978.1379
40 15 a 159524.845 1200.06 47191.1934 135,778.161 1200 84630.555
40 15 b 154688.16 1200.07 41516.1214 139,717.219 1200 91314.4459
40 15 c 156508.747 1200.01 59948.6305 132,638.426 1200 75262.5445
40 15 d 155834.933 1200.01 78912.2935 136,892.519 1200.01 77414.0106
40 15 e 165718.822 1200 50292.6871 137,059.666 1200.01 80641.5541
40 20 a 263795.958 1200.16 211.213 144,504.175 1200 71539.575
40 20 b 364351.538 1200.18 218.9343 143,862.081 1200 80572.0252
40 20 c 665331.391 1200.13 251.3194 141,347.44 1200.01 78199.6221
40 20 d 167943.89 1200.09 62785.1164 139,146.865 1200 78033.3746
40 20 e 328050.578 1200.11 43450.2513 143,208.561 1200 91381.3884
50 2 a 44038.524 1200.01 18269.491 41,790.384 1200 21189.9737
50 2 b 46,906.656 1200 27594.1435 46,906.656 1200 37987.115
50 2 c 47492.538 1200.01 25402.1795 46,954.195 1200 31348.1609
50 2 d 53796.88 1200 40521.8152 52,412.181 409.8 52412.181
50 2 e 42740.866 1200.01 28068.8009 42,621.504 1200 29362.3917
50 3 a 70573.677 1200 39418.9706 68,320.363 1200 44638.5789
50 3 b 72583.313 1200 30825.6339 72,180.859 1200 37424.6444
50 3 c 68545.294 1200 38233.0965 66,476.403 1200 43243.2062
50 3 d 74127.786 1200 36994.3169 71,001.682 1200 40968.0831
50 3 e 81466.739 1200.01 45967.7125 68,197.434 1200 50919.4102
50 5 a 111911.158 1200.02 42715.6481 99,117.746 1200.01 46229.1332
50 5 b 104810.22 1200 52223.4928 94,336.443 1200 58936.9436
50 5 c 114922.188 1200.01 44401.6886 99,118.254 1200 46066.0562
50 5 d 114144.67 1200.01 40758.0377 96,798.509 1200 57305.9588
50 5 e 117526.411 1200.01 38050.4846 103,507.577 1200 46385.7182
50 10 a 131644.913 1200.03 35484.0848 125,454.434 1200 64063.8249
50 10 b 136810.053 1200.04 53466.0705 129,899.7 1200 65476.065

123
119 Page 64 of 72 B. Nikolic et al.

Table 23 continued
n m T ype FdSdS New

r t[s] LB r t[s] LB

50 10 c 140229.245 1200.03 55941.5775 126,603.906 1200.01 57922.9736

50 10 d 141722.013 1200.05 46575.4685 126,527.477 1200 56442.7203
50 10 e 149685.648 1200 39651.6763 125,686.149 1200 66127.6289
50 15 a 157897.026 1200.06 17650.2299 136,204.826 1200.01 64950.637
50 15 b 156728.626 1200.08 36554.0247 145,269.872 1200 72246.782
50 15 c 168341.93 1200.08 40310.7899 144,227.83 1200 62037.1268
50 15 d 157074.072 1200.08 33230.2372 145,081.588 1200.01 59073.2001
50 15 e 169580.945 1200.09 41192.3978 138,135.247 1200 64374.6512
50 20 a 879304.004 1200.07 19223.5462 151,382.038 1200 60100.8581
50 20 b 230964.8 1200.17 31656.5855 155,334.547 1200 72526.4574
50 20 c 263435.795 1200.11 74.3912 151,001.377 1200.01 62756.1479
50 20 d 907439.851 1200.14 32478.7937 150,532.516 1200.01 59561.1838
50 20 e 482080.833 1200.23 175.3652 151,836.601 1200.01 82636.7756

123
New mixed-integer linear programming model... Page 65 of 72 119

Table 24 Results for n = 100, 300, 500 and k = 30

n m T ype FdSdS New

r t[s] LB r t[s] LB

100 2 a 33795.249 1200 0 17,247.181 1200.01 0

100 2 b 30679.797 1200 0 22,706.532 1200 0
100 2 c 27593.52 1200 0 16,592.208 1200 0
100 2 d 31750.464 1200.01 0 16,975.308 1200 0
100 2 e 31028.727 1200 0 11,915.383 1200 0
100 3 a 69926.921 1200.02 0 44,669.574 1200 0
100 3 b 65166.53 1200 0 42,615.909 1200 0
100 3 c 55469.589 1200 0 48,864.373 1200.01 0
100 3 d 67757.193 1200.02 0 42,710.83 1200 0
100 3 e 77455.808 1200.01 0 48,217.645 1200.01 0
100 5 a 113341.855 1200.01 0 73,532.576 1200 0
100 5 b 104807.285 1200.03 0 84,570.21 1200.01 0
100 5 c 114360.583 1200.01 0 79,415.537 1200 0
100 5 d 98944.047 1200.02 0 87,010.239 1200.01 0
100 5 e 112734.198 1200.03 0 85,377.188 1200 0
100 10 a 168138.262 1200.02 0 125,633.84 1200.01 0
100 10 b 148209.045 1200.04 0 136,909.213 1200.01 13213.7606
100 10 c 163707.598 1200.3 0 127,955.426 1200 0
100 10 d 147351.055 1200.06 0 40,420.74 1200.01 0
100 10 e 159186.162 1200.06 0 133,132.31 1200.01 12729.3958
100 15 a 324834.121 1200.18 8913.381 161,377.343 1200.01 20161.7951
100 15 b 268465.224 1200.18 10017.1618 159,666.805 1200.01 18688.9421

123
119 Page 66 of 72 B. Nikolic et al.

Table 24 continued
n m T ype FdSdS New

r t[s] LB r t[s] LB

100 15 c 314493.485 1200.07 5947.2333 173,343.972 1200.01 20312.0173

100 15 d 335394.086 1200.15 7322.3815 155,478.361 1200.02 15876.5348
100 15 e 198293.901 1200.08 6440.8593 173,638.042 1200.01 15290.2138
100 20 a 293435.513 1200.16 1727.9619 180,746.374 1200.02 14835.6736
100 20 b 319472.361 1200.2 0 172,987.098 1200.01 20471.3104
100 20 c 295039.109 1200.23 0 179,049.591 1200 15116.9964
100 20 d 328222.772 1200.22 0 185,212.95 1200.01 20865.6854
100 20 e 381863.596 1200.26 13818.1073 191,371.739 1200.01 25551.6852
300 2 a 30073.813 1200.02 0 17,603.693 1201.89 0
300 2 b 23665.193 1200.01 0 13,573.363 1200.02 0
300 2 c 24118.019 1200.02 0 15,385.168 1201.81 0
300 2 d 22681.587 1200.01 0 11,513.436 1200.54 0
300 2 e 22469.267 1200 0 19,313.25 1200.01 0
300 3 a 60110.288 1200.01 0 45,574.891 1200.01 0
300 3 b 50730.267 1200.05 0 34,789.825 1200.02 0
300 3 c 53790.266 1200.02 0 37,026.153 1200.01 0
300 3 d 58664.787 1200.03 0 35,064.137 1200.01 0
300 3 e 48186.639 1200.01 0 44,247.806 1200.02 0
300 5 a 120988.226 1200.39 0 72,962.087 1200 0
300 5 b 119236.781 1200.46 0 88,254.853 1200.01 0
300 5 c 111963.587 1200.41 0 70,068.396 1200.01 0
300 5 d 122842.843 1208.72 0 70,080.3 1200.01 0
300 5 e 134523.856 1200.54 0 78,316.321 1200.01 0
300 10 a 176436.976 1200.33 0 142,530.183 1200.01 0
300 10 b 188365.144 1200.19 0 141,577.266 1200.01 0
300 10 c 189743.993 1200.08 0 154,596.745 1200 0
300 10 d 139,233.278 1200.17 0 148433.74 1200.01 0
300 10 e 172531.112 1200.1 0 150,829.842 1200.01 0
300 15 a 218408.604 1200.21 0 190,766.322 1200.02 0
300 15 b 218319.221 1200.16 0 171,087.529 1200.01 0
300 15 c 312960.063 1200.1 0 197,420.8 1200.02 0
300 15 d 253433.959 1200.48 0 194,624.941 1200.01 0
300 15 e 258128.965 1200.35 0 182,665.387 1200.03 0
300 20 a 344294.436 1200.69 0 203,788.99 1200.03 0
300 20 b 403083.812 1200.76 0 177,420.249 1200.02 0
300 20 c 346626.593 1200.26 0 202,894.119 1200.02 0
300 20 d 368032.543 1200.87 0 231,428.932 1200.04 0
300 20 e 377355.084 1200.74 0 209,873.225 1200.01 0

123
New mixed-integer linear programming model... Page 67 of 72 119

Table 24 continued
n m T ype FdSdS New

r t[s] LB r t[s] LB

500 2 a 20460.434 1200.02 0 14,424.059 1200.31 0

500 2 b 20275.721 1200.02 0 14,304.523 1200.03 0
500 2 c 16139.057 1200.03 0 13,738.81 1200.12 0
500 2 d 19515.743 1200.05 0 13,524.126 1200.03 0
500 2 e 17,783.455 1200.06 0 21848.822 1200.03 0
500 3 a 91905.565 1200.38 0 30,330.474 1200.02 0
500 3 b 69355.91 1200.35 0 37,089.192 1200.02 0
500 3 c 62645.769 1200.95 0 52,018.446 1200.02 0
500 3 d 70460.765 1200.36 0 28,417.351 1200.03 0
500 3 e 75836.937 1200.39 0 53,684.861 1200.04 0
500 5 a 103273.871 1200.93 0 68,628.48 1200.01 0
500 5 b 109283.714 1200.63 0 76,051.29 1200.01 0
500 5 c 126480.115 1200.57 0 85,317.895 1200.01 0
500 5 d 108885.954 1200.02 0 70,931.701 1200.02 0
500 5 e 122069.934 1200.02 0 69,240.317 1200.01 0
500 10 a 219468.074 1200.43 0 143,356.292 1200 0
500 10 b 175880.532 1210.71 0 138,372.188 1200.01 0
500 10 c 167988.004 1203.35 0 130,006.924 1200.01 0
500 10 d 200656.406 1200.08 0 147,517.656 1200.01 0
500 10 e 166645.678 1203.92 0 147,617.121 1200.01 0
500 15 a 266014.437 1200.55 0 191,771.568 1200.04 0
500 15 b 296581.767 1200.14 0 165,587.041 1200.03 0
500 15 c 275012.367 1200.56 0 182,125.167 1200.04 0
500 15 d 358607.455 1200.58 0 208,769.829 1200.03 0
500 15 e 291411.457 1200.56 0 176,933.188 1200.04 0
500 20 a 404633.233 1200.81 0 225,641.363 1200.17 0
500 20 b 418527.599 1200.19 0 223,993.495 1200.01 0
500 20 c 331494.553 1200.69 0 221,685.662 1200.17 0
500 20 d 401011.827 1200.81 0 198,875.414 1200.03 0
500 20 e 461161.234 1200.77 0 224,112.052 1200.18 0

Table 25 Results for n = 200, 400 and k = 30

n m T ype FdSdS New

r t[s] LB r t[s] LB

200 2 a 27142.123 1200 0 20,055.708 1200.58 0

200 2 b 31561.089 1200 0 18,009.89 1200.01 0
200 2 c 18245.553 1200.01 0 16,716.786 1200.01 0
200 2 d 21295.207 1200.01 0 19,149.396 1200.01 0

123
119 Page 68 of 72 B. Nikolic et al.

Table 25 continued
n m T ype FdSdS New

r t[s] LB r t[s] LB

200 2 e 29751.618 1200.03 0 23,481.054 1200.01 0

200 3 a 60886.156 1200.01 0 47,099.942 1200.01 0
200 3 b 59479.384 1200.02 0 56,987.7 1200.01 0
200 3 c 68067.445 1200.01 0 38,314.194 1200.34 0
200 3 d 48720.442 1200.03 0 37,484.179 1200.01 0
200 3 e 64861.358 1200.02 0 47,069.376 1200.01 0
200 5 a 101893.955 1200.02 0 87,281.3 1200.01 0
200 5 b 91939.812 1200.05 0 72,738.565 1200.01 0
200 5 c 102521.436 1200 0 92,128.657 1200 0
200 5 d 96670.849 1200.05 0 79,368.219 1200 0
200 5 e 111026.462 1200.04 0 75,587.944 1200.01 0
200 10 a 186456.854 1200.23 0 129,927.382 1200.01 0
200 10 b 191897.337 1200.01 0 143,435.808 1200 0
200 10 c 144,881.763 1200.07 0 152779.562 1200.01 0
200 10 d 141149.666 1200.22 0 126,730.246 1200.01 0
200 10 e 208143.886 1200.05 0 139,525.216 1200.02 0
200 15 a 209492.081 1200.06 0 176,792.552 1200.03 0
200 15 b 244182.356 1200.03 0 186,001.032 1200.01 0
200 15 c 237833.014 1200.14 0 172,702.002 1200.01 0
200 15 d 233225.946 1200.07 0 169,232.314 1200.01 0
200 15 e 299695.466 1200.15 0 191,672.537 1200.03 0
200 20 a 370435.96 1200.74 0 209,022.086 1200.02 0
200 20 b 378449.861 1200.29 0 181,388.343 1200.03 0
200 20 c 377734.108 1200.3 0 202,153.152 1200.04 0
200 20 d 335722.962 1200.19 0 192,179.384 1200.01 0
200 20 e 350278.73 1200.24 0 198,703.202 1200.01 0
400 2 a 17903.731 1200.05 0 13,384.939 1200.11 0
400 2 b 17835.496 1200.02 0 15,761.954 1200.17 0
400 2 c 24408.702 1200.03 0 23,345.858 1200.03 0
400 2 d 19079.41 1200.03 0 14,328.506 1200.04 0
400 2 e 32164.011 1200.03 0 16,767.723 1200.87 0
400 3 a 76500.831 1200.03 0 47,521.799 1200.01 0
400 3 b 69897.618 1200.04 0 43,752.644 1200.02 0
400 3 c 86046.87 1200.37 0 49,587.578 1200.02 0
400 3 d 74116.867 1200.02 0 43,045.61 1200.02 0
400 3 e 76396.694 1200.29 0 46,378.14 1200.02 0
400 5 a 102350.709 1200.63 0 90,284.267 1200.01 0
400 5 b 101424.225 1200.54 0 79,370.322 1200.01 0

123
New mixed-integer linear programming model... Page 69 of 72 119

Table 25 continued
n m T ype FdSdS New

r t[s] LB r t[s] LB

400 5 c 131755.977 1200.27 0 62,203.051 1200.03 0

400 5 d 93173.214 1200.46 0 71,567.544 1200.01 0
400 5 e 124530.206 1200.49 0 82,230.65 1200.01 0
400 10 a 157416.293 1202.32 0 154,393.888 1200.01 0
400 10 b 164721.96 1206.33 0 141,098.564 1200.01 0
400 10 c 192125.462 1200.09 0 141,719.199 1200.01 0
400 10 d 182752.591 1200.04 0 129,064.43 1200.01 0
400 10 e 147,370.509 1200.23 0 153404.239 1200.01 0
400 15 a 186,635.482 1200.61 0 198905.838 1200.03 0
400 15 b 262613.021 1200.51 0 199,343.226 1200.03 0
400 15 c 293718.029 1200.17 0 171,842.428 1200.03 0
400 15 d 281342.361 1200.11 0 213,618.833 1200.03 0
400 15 e 219797.407 1200.77 0 192,171.991 1200.02 0
400 20 a 359727.164 1201.27 0 195,824.004 1200.05 0
400 20 b 397111.02 1201.1 0 231,497.352 1200.04 0
400 20 c 386961.468 1201.03 0 215,117.468 1200.05 0
400 20 d 345029.688 1201 0 222,724.931 1200.01 0
400 20 e 406917.962 1201.5 0 235,989.533 1200.06 0

Appendix B: Proof
 m
In this appendix, we show that for a given set of vectors S ⊆ R+ , and arbitrary integer
k ∈ {2, . . . , |S|}, the objective function r of the MDMWNPP does not reach a minimum
value for the k–decomposition which contains an empty set. More precisely, we prove that
for each such k−decomposition of set S there is a corresponding k− partition of set S with
smaller value of the objective function r .
In accordance with the notation introduced in the definition of the model NEW, we use
the following abbreviations: if S = {S j | j ∈ {1, . . . , k}} is a k−decomposition of set S,
then
⎛ ⎞

yl (S ) = max ⎝ vil ⎠ , ∀l ∈ {1, . . . , m},
j∈{1,...,k}
vi ∈S j
⎛ ⎞

z l (S ) = min ⎝ vil ⎠ , ∀l ∈ {1, . . . , m},
j∈{1,...,k}
vi ∈S j

r (S ) = max {yl (S ) − zl (S )} .
l∈{1,...,m}
 m
Lemma 1 Let m, n ≥ 2 be a fixed integers, and S = {vi ∈ R+ | i ∈ {1, 2, . . . , n}} is
a given set of vectors. For an arbitrary integer k ∈ {2, . . . , n} and any k−decomposition
S of set S which contains exactly one empty set exists k–partition S of set S such that
r (S ) < r (S ).

123
119 Page 70 of 72 B. Nikolic et al.

Proof Let k ∈ {2, . . . , n} be an arbitrary integer and let S be an arbitrary k–decomposition of

set S which contains exactly one empty set. Then S = {S1 , . . . , Sk−1 }∪{∅}, for some (k −1)–
partition {S1 , . . . , Sk−1 } of set S. If we adopt convention vil = 0, ∀l ∈ {1, . . . , m}, then
vi ∈∅
it is obvious fact that zl (S ) = 0, ∀l ∈ {1, . . . , m}, which implies r (S ) = max {yl (S )}.
l∈{1,...,m}
Since k ≤ n, we can choose index j0 ∈ {1, . . . , k − 1} such that set S j0 contains at least
two vectors. If we pick a vector vi0 ∈ S j0 for some i 0 ∈ {1, 2, . . . , n}, then we can define
   
non-empty sets S j0 := vi0 and S j0 := S j0 \ vi0 . Using this sets, we can construct a
k–partition
   
S := S j | j ∈ {1, . . . , k − 1} \ { j0 } ∪ S j0 , S j0

of set S. To complete the proof, it is sufficient to show that partition S has the required
property. Every member of the family S contains at least one vector   and all coordinates
of all vectors are positive real numbers, which implies that zl S > 0 holds for any
 
l ∈ {1, . . . , m}. From this observation and definition of yl S , we get
         
r S = max yl S − zl S < max yl S
l∈{1,...,m} l∈{1,...,m}
⎧ ⎧ ⎫ ⎫
⎪
⎨ ⎨ ⎬  ⎪
⎬

= max max vil , vil , vil
⎩ j∈{1,...,k−1}\{ j0 } ⎩vi ∈S j ⎭
l∈{1,...,m} ⎪ ⎪
⎭
vi ∈S j vi ∈S j
0 0
⎧ ⎧ ⎫ ⎛ ⎞ ⎫
⎨ ⎨ ⎬  ⎬
= max max vil , vi0 l , ⎝ vil ⎠ − vi0 l
l∈{1,...,m} ⎩ j∈{1,...,k−1}\{ j0 } ⎩ ⎭ ⎭
vi ∈S j vi ∈S j0

For each index l ∈ {1, . . . , m}, we get the following inequalities

⎧ ⎫ ⎧ ⎫
⎨ ⎬ ⎨ ⎬
max vil ≤ max vil = yl (S ) ,
j∈{1,...,k−1}\{ j0 } ⎩ ⎭ j∈{1,...,k−1} ⎩ ⎭
vi ∈S j vi ∈S j
⎧ ⎫
 ⎨ ⎬
vi0 l ≤ vil ≤ max vil = yl (S ) ,
j∈{1,...,k−1} ⎩ ⎭
vi ∈S j0 vi ∈S j
⎛ ⎞ ⎧ ⎫
  ⎨ ⎬
⎝ vil ⎠ − vi0 l < vil ≤ max vil = yl (S ) .
j∈{1,...,k−1} ⎩ ⎭
vi ∈S j0 vi ∈S j0 vi ∈S j

This means that inequality yl (S ) ≤ yl (S ) holds for each index l ∈ {1, . . . , m}. From this
fact and the previous observations, we conclude r (S ) < r (S ).

Now, we generalize the claim from the previous lemma to the case of k-decompositions
of set S with at least one empty set.
 m
Theorem 1 Let m, n ≥ 2 be a fixed integers, and S = {vi ∈ R+ | i ∈ {1, . . . , n}} is
a given set of vectors. For an arbitrary integer k ∈ {2, . . . , n} and any k–decomposition
S of set S which contains at least one empty sets exists k–partition S of set S such that
r (S ) < r (S ).

123
New mixed-integer linear programming model... Page 71 of 72 119

Proof Let k ∈ {2, . . . , n} be an arbitrary integer and let S be an arbitrary k–decomposition of

set S which contains at least one empty sets. If the family S contains exactly q ∈ {1, . . . , k−1}
empty sets, then exists a (k − q)–partition S0 of set S such that S = S0 ∪ {∅, . . . , ∅}. The
  
q
family S0 ∪ {∅} is a (k − q + 1)-decomposition of set S and, since k − q < k ≤ n, we
can use the previous lema to construct a (k − q + 1)–partition S1 of set S with property
r (S1 ) < r (S0 ∪ {∅}). If k − q + 1 < k (i.e., if q > 1), then, from the construction of the
family S1 , follows that the family S1 ∪{∅} is a (k−q +2)-decomposition of set S with property
r (S1 ∪ {∅}) ≤ r (S0 ∪ {∅, ∅}), and, since k − q + 1 < n, we can again apply the previous
lemma to construct a (k − q + 2)-partition S2 of set S with property r (S2 ) < r (S1 ∪ {∅}).
In this way, by successive application of the previous lemma, we construct a finite sequence
S1 , S2 , . . . , Sq of families with the following properties:

(i) For p ∈ {1, . . . , q}, the family S p is a (k − q + p) − partition of set S.

(ii) For p ∈ {1, . . . , q}, r (S p ) < r (S p−1 ∪ {∅}).
(iii) For p ∈ {1, . . . , q − 1}, r (S p ∪ {∅}) ≤ r (S p−1 ∪ {∅, ∅}).

To complete the proof, it is sufficient to show that family S := Sq has the required property.
Indeed, the condition (i) implies that S is a k-partition of set S, while applying conditions
(ii) and (iii), we conclude
r (S ) < r (Sq−1 ∪ {∅}) ≤ r (Sq−2 ∪ {∅, ∅}) ≤ . . . ≤ r (S0 ∪ {∅, . . . , ∅}) = r (S ).
  
q

The consequence of the previous theorem is that from a standpoint of minimum of objective
function of the MDMWNPP it makes no difference in the definition of this problem if we
require that a family {S j | j ∈ {1, 2, . . . , k}} is a k–decomposition or a k–partition of set S,
m
as long as the condition S ⊆ R+ is fulfilled. This resolves a potential ambiguity which
may arise in the definition of the MDMWNPP, since all the published papers regarding
this problem used the same starting sets of instances for which the mentioned condition is
satisfied.

References
Coffman EG, Lueker GS (1991) Probabilistic analysis of packing and partitioning algorithms. Wiley, Ams-
terdam
Faria AF, de Souza SR, de Sá EM (2020) A mixed-integer linear programming model to solve the multidi-
mensional multi-way number partitioning problem. Comput Oper Res 127:105133
Fuksz L, Pop P, Zelina I (2013) Heuristic algorithms for solving the bi-dimensional two-way number parti-
tioning problem. Stud Univ Babes-Bolyai Ser Inf 58(3):17–28
Hacibeyoglu M, Alaykiran K, Acilar AM, Tongur V, Ulker E (2018) A comparative analysis of metaheuristic
approaches for multidimensional two-way number partitioning problem. Arab J Sci Eng 43(12):7499–
7520
Hacibeyoglu M, Tongur V, Alaykiran K (2014) Solving the bi-dimensional two-way number partitioning prob-
lem with heuristic algorithms. In: 2014 IEEE 8th international conference on application of information
and communication technologies (AICT), pp 1–5, IEEE
Hayes B (2002) Computing science: the easiest hard problem. Am Sci 90(2):113–117
Karp RM (1975) On the computational complexity of combinatorial problems. Networks 5(1):45–68
Kojić J (2010) Integer linear programming model for multidimensional two-way number partitioning problem.
Comput Math Appl 60(8):2302–2308

123
119 Page 72 of 72 B. Nikolic et al.

Korf RE (2009) Multi-way number partitioning. IJCAI 9:538–543

Korf, RE, Schreiber, EL, Moffitt MD (2014) Optimal sequential multi-way number partitioning. In: ISAIM
Kratica J, Kojić J, Savić A (2014) Two metaheuristic approaches for solving multidimensional two-way number
partitioning problem. Comput Oper Res 46:59–68
Ercegovac DM, Potkonjak M (1999) Low-power behavioral synthesis optimization using multiple precision
arithmetic. In: Proceedings of the 36th annual ACM/IEEE design automation conference, pp 568–573.
Association for Computing Machinery
Mertens S (2006) The easiest hard problem: number partitioning. Comput Complex Stat Phys 125(2):125–139
Moffitt MD (2013) Search strategies for optimal multi-way number partitioning. In: 23rd International Joint
Conference on Artificial Intelligence
Pohlert T (2014) The pairwise multiple comparison of mean ranks package (pmcmr). R package 27(2020):10
Pop PC, Matei O (2013) A genetic algorithm approach for the multidimensional two-way number partitioning
problem. In: International conference on learning and intelligent optimization, pp 81–86. Springer
Pop PC, Matei O (2013) A memetic algorithm approach for solving the multidimensional multi-way number
partitioning problem. Appl Math Model 37(22):9191–9202
Merkle R, Hellman M (1978) Hiding information and signatures in trapdoor knapsacks. IEEE Trans Inf Theory
24(5):525–530
Rodriguez FJ, Glover F, García-Martínez C, Martí R, Lozano M (2017) Grasp with exterior path-relinking and
restricted local search for the multidimensional two-way number partitioning problem. Comput Oper
Res 78:243–254
Santucci V, Baioletti M, Di Bari G, Milani A (2019) A binary algebraic differential evolution for the multidi-
mensional two-way number partitioning problem. In: European conference on evolutionary computation
in combinatorial optimization (Part of EvoStar), pp 17–32. Springer

Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional affiliations.

123