mathematics

Article
Packaging Process Optimization in Multihead Weighers with
Double-Layered Upright and Diagonal Systems
Rafael García-Jiménez 1, * , J. Carlos García-Díaz 2 and Alexander D. Pulido-Rojano 3

1 Department of Exact Sciences, Universidad Simón Bolívar, Barranquilla 080020, Colombia

2 Department of Applied Statistics and Operations Research and Quality, Universitat Politècnica de València,
46022 Valencia, Spain; juagardi@eio.upv.es
3 Grupo de Consultoría e Innovación JJ&N S.A.S., Department of Industrial Engineering,
Universidad Simón Bolívar, Barranquilla 080020, Colombia; apulido3@unisimonbolivar.edu.co
* Correspondence: rgarcia1@unisimonbolivar.edu.co

Abstract: In multihead weighers, packaging processes seek to find the best combination of passage
hoppers whose product content provides a total package weight as close as possible to its (nominal)
label weight. The weighing hoppers arranged in these machines dispense the product quantity that
each package contains through computer algorithms designed and executed for this purpose. For
its part, in the packaging process for double-layered multihead weighers, all hoppers are arranged
in two levels. The first layer comprises a group of weighing hoppers, and the second comprises a
set of booster hoppers placed uprightly or diagonally to each weighing hopper based on design of
the machine. In both processes, the initial machine configuration is the same; however, the hopper
selection algorithm works differently. This paper proposes a new packaging process optimization





 algorithm for double-layer upright and diagonal machines, wherein the hopper subset combined
Citation: García-Jiménez, R.;
has previously been defined, and the packaging weight is expressed as actual values. As part
García-Díaz, J.C.; Pulido-Rojano, A.D. of its validation, product filling strategies were implemented for weighing hoppers to assess the
Packaging Process Optimization in algorithm in different scenarios. Results from the process performance metrics prove that the new
Multihead Weighers with algorithm improves processes by reducing variability. In addition, results reveal that some machine
Double-Layered Upright and configurations were also able to improve their operation.
Diagonal Systems. Mathematics 2021,
9, 1039. https://doi.org/10.3390/ Keywords: multihead machines; packaging processes; double layer; hoppers filling strategies; optimization
math9091039

Academic Editor: Nicu Bizon

1. Introduction
Received: 18 March 2021
Accepted: 22 April 2021
Multihead weighers were first used in the 1970s and, since then, they have been
Published: 4 May 2021
used to improve industrial process performance. Currently, they are used in different
packaging processes for several products, regardless of their shape or size, and have become
Publisher’s Note: MDPI stays neutral
particularly essential in the food industry [1,2]. Multihead weighers play a fundamental
with regard to jurisdictional claims in
role in the industrial process given the accuracy and speed they provide to the packaging
published maps and institutional affil- process. Once programmed, they can automatically package several food industry products,
iations. including products with heterogeneous characteristics. This ensures compliance with
current regulatory standards while reducing production costs by maintaining an efficient,
high-volume product flow and preserving process quality. Still, one of the most prominent
process requirements is that package contents must match the weight printed on its label.
Copyright: © 2021 by the authors.
This becomes one of the main issues that must be addressed to guarantee high-quality
Licensee MDPI, Basel, Switzerland.
standards in the final product, with variability being directly related to the quality of the
This article is an open access article
process [3]. In this sense, authors such as Taguchi [4] claim that product quality depends
distributed under the terms and on how close its quality features are to their nominal value, in such a way that everything
conditions of the Creative Commons that deviates from said nominal value is considered as a loss to society. Hence, quality in
Attribution (CC BY) license (https:// the packaging process may be achieved by reducing variability around the target weight
creativecommons.org/licenses/by/ (T) [5]. For these purposes, processes must be streamlined trying to guarantee that package
4.0/). content weights are as close as possible to the weight specified on the label. Optimization

Mathematics 2021, 9, 1039. https://doi.org/10.3390/math9091039 https://www.mdpi.com/journal/mathematics

 Mathematics 2021, 9, x FOR PEER REVIEW 2 of 21

Mathematics 2021, 9, 1039 variability around the target weight (𝑇) [5]. For these purposes, processes must be
2 of 20
streamlined trying to guarantee that package content weights are as close as possible to
the weight specified on the label. Optimization techniques, new algorithms, and
experiments’ design can be used as quality engineering tools for the improvement and
techniques, new algorithms, and experiments’ design can be used as quality engineering
optimization of industrial processes [6].
tools for the improvement and optimization of industrial processes [6].
Multihead weighing systems are mainly composed of a vibrating feeder system, 𝑛
Multihead weighing systems are mainly composed of a vibrating feeder system,
feeding hoppers (𝐻 ), 𝑛 weighing hoppers (𝐻 ), and a discharge conduit to a packaging
n feeding hoppers (HF ), n weighing hoppers ( HW ), and a discharge conduit to a packaging
machine. In a traditional single-layer machine (without booster hoppers), the internal
machine. In a traditional single-layer machine (without booster hoppers), the internal
algorithm combines all the 𝑛 weighing hoppers in the machine to form a subgroup 𝐻 of
algorithm combines all the n weighing hoppers in the machine to form a subgroup H of k
𝑘 hoppers and then selects a subset 𝐻′ ∈ H whose sum of weights is the closest to the
hoppers and then selects a subset H 0 ∈ HW whose sum of weights is the closest to the target
target weight 𝑇 of the package. In this sense, package content weights are obtained by
weight T of the package. In this sense, package content weights are obtained by combining
combining a subgroup of 𝑘 hoppers from the 𝑛 weighing hoppers that the machine
a subgroup of k hoppers from the n weighing hoppers that the machine contains, with
contains, with 𝑘 ≤ 𝑛. However, double-layered (upright and diagonal) weighers contain
k ≤ n. However, double-layered (upright and diagonal) weighers contain an additional
an additional
layer of hopperslayer
andofuse
hoppers andcombination
a larger use a larger combination
of hoppers toofreachhoppers to reach
the target the target
weight. In
weight. In these double-layered machines, each of the 𝑛 weighing hoppers
these double-layered machines, each of the n weighing hoppers (HW ) has a booster hopper (𝐻 ) has a
booster
(H hopper (𝐻 ) underneath, in which a single product ration can be stored, while the
B ) underneath, in which a single product ration can be stored, while the weighing
weighing
hoppers receivehoppers receive
a new a new
product product
ration from theration from hoppers
feeding the feeding(HF ),hoppers ( 𝐻 ), in
as illustrated as
illustrated in Figure 1. For each package, all the hoppers must contain
Figure 1. For each package, all the hoppers must contain a certain product quantity, and a certain product
quantity,
the machine and thecombine
must machinek must combine
hoppers 𝑘 hoppers
of the 2n so that the the 2𝑛 must
of system so that
form thea system
subset Hmust
0 of
form a subset 𝐻 of 𝑘 hoppers. For double-layered upright 0 systems,
k hoppers. For double-layered upright systems, this subset H can be formed with booster this subset 𝐻 can
be formed with booster hoppers or weighing hoppers as long as its corresponding
hoppers or weighing hoppers as long as its corresponding booster hopper is part of the k booster
hopper is
hoppers, part for
while, of the 𝑘 hoppers, while,
thedouble-layered for the
diagonal double-layered
systems, a weighing diagonal
and booster systems,
hoppera
weighing and booster hopper cannot be selected
cannot be selected simultaneously for the same k-nuple. simultaneously for the same k-nuple.

Figure1.1.Weighing
Figure Weighingand
anddischarge
dischargeconfiguration
configurationon
ondouble-layered
double-layeredmachines.
machines.

This
Thisresearch
researchstudy proposes
study a new
proposes a optimization algorithm
new optimization for double-layered
algorithm upright
for double-layered
and diagonal machines that seeks to select a subset H 0 of k hoppers whose total weights
upright and diagonal machines that seeks to select a subset 𝐻 of 𝑘 hoppers whose total
are greaterare
weights and closer and
greater to thecloser
targettoweight T. Theweight
the target algorithm was algorithm
𝑇 . The validated through a case
was validated
study
through a case study for actual products, wherein the performance characteristics were
for actual products, wherein the performance characteristics of the process of the
measured in different scenarios. In numerical experiments, filling strategies designed
process were measured in different scenarios. In numerical experiments, filling strategies and
validated
designed forandthree weighing
validated forhoppers in previous
three weighing studiesinwere
hoppers tested.studies
previous Likewise,
weredifferent
tested.
values from previously established k combined hoppers were tested in each filling strategy.
Likewise, different values from previously established 𝑘 combined hoppers were tested
This article is structured as follows: Section 2 presents the state of the art within
in each filling strategy.
the multihead machine packaging process field. In Section 3, the weighing process for
This article is structured as follows: Section 2 presents the state of the art within the
double-layered upright and diagonal multihead weighers is described. Section 4 discusses
multihead machine packaging process field. In Section 3, the weighing process for double-
the different hopper-filling strategies used in the process. Then, Section 5 explains the
layered upright and diagonal multihead weighers is described. Section 4 discusses the
optimization model and the packaging algorithm proposed for double-layered weighers.
Section 6 lists the results and provides a preliminary assessment of the numerical experi-
ments. Next, Section 7 contains the results from the experimental design for the different
Mathematics 2021, 9, 1039 3 of 20

process scenarios evaluated. Finally, Section 8 discussion and Section 9 elucidate our
conclusions from this research study.

2. Background
The documented research studies of multihead machines propose diverse solutions
for packaging issues. Different algorithms and solution and configuration models have
been presented for these processes. For example, some authors propose an optimal scheme
for determining ideal operation times for vibrating linear feeders in multihead weighers.
In their corresponding studies, they found that the least-squares method can be used
to reduce variability against fixed operating times [7]. A bit-operation-based weighing
algorithm succeeded in proving that operation times can be reduced and that the output
weight from uneven hoppers is closer to the desired target weight [8]. In other research
studies, the product residence time in hoppers was introduced as a second objective.
Hence, the packaging problem was formulated as a bicriteria approximation issue for
discrete weights, and a dynamic programming algorithm was proposed for its solution.
This algorithm aimed to reduce the maximum duration of the product in the packaging
system [9] while assuring that the total weight of each package was as close as possible to
its target weight. Similar work was approached for double-layered upright and diagonal
machines for discrete weights in the hoppers, using a bicriteria dynamic programming
algorithm for these machines [10]. Subsequently, the bicriteria model proposed by [9] was
ultimately improved through algorithms designed to reduce execution times, which makes
the packaging process more efficient [11]. Later, these theories were extended to duplex and
quasi-duplex machines [12]. In the same sense, heuristic algorithms have been proposed to
find better results in the bicriteria packaging process, targeting label weights, and priority
orders [13,14]. Likewise, several optimization algorithms have been proposed to determine
the optimum product flow rate for radial feeders, seeking to minimize expected production
costs per each “compliant” package within a fixed period of time. In this study, the response
surface methodology identified conditions of minimal process variability and lower costs,
compared to an industrial solution [15]. A heuristic optimization model based on a detailed
characterization of what constitutes a near-optimal solution to the multihead weigher
configuration problem has also been proposed. This model reduced hopper combination
response times according to each package weight, in addition to finding the right hopper
feeding points to minimize the mean square error for package weights [16]. Statistical
control of the packaging process has also been addressed. For example, several modified
control charts have been developed and studied to monitor and control the package
production process [17]. Finally, new bio-objective approaches have been developed for
the optimization of actual package weights [18,19], incorporating package target weights
and the priority associated with the product discharge times reported by hoppers into the
model. The innovation proposed by these approaches is that the relative importance of
the objectives is managed and dynamically adjusted in each packaging operation, thus
determining the best operation conditions for each process [20]. These approaches are
extremely useful for packaging fresh or frozen products. As evidenced, hitherto, no
scientific content has been disclosed regarding diagonal and upright machines considering
actual weights in hoppers with previously established k values. Therefore, this study
becomes quite relevant as it proposes a new optimization algorithm for double-layered
upright machines that considers actual weights and preestablished k values.

3. Weighing Process on Double-Layered Upright and Diagonal Weighers

The packaging process in double-layered multihead weighers consists of selecting a
subset H 0 from the set H of the 2n machine hoppers to produce each package. Initially, the
product is distributed through the vibrating channels to the feeding hoppers, Then, it is
discharged to the HWi weighing hoppers (i = 1, 2, 3, . . . n), where the product is weighed.
In double-layered upright and double-layered diagonal weighers, each of the weighing
hoppers has a booster hopper, in which the content of the weighed product can be stored.
 3. Weighing Process on Double-Layered Upright and Diagonal Weighers
The packaging process in double-layered multihead weighers consists of selecting a
subset 𝐻 from the set 𝐻 of the 2𝑛 machine hoppers to produce each package. Initially,
Mathematics 2021, 9, 1039
the product is distributed through the vibrating channels to the feeding hoppers, Then, it
4 of 20
is discharged to the 𝐻 weighing hoppers (𝑖 = 1,2,3, … 𝑛), where the product is weighed.
In double-layered upright and double-layered diagonal weighers, each of the weighing
hoppers has a booster hopper, in which the content of the weighed product can be stored.
For this reason, once the product is weighed, it is discharged into the corresponding H 𝐻Bj
booster hoppers (j (𝑗 = 𝑛
n + 1, 𝑛n + 2, …. . .2𝑛),
2n), and
and the
the weighing
weighing hopper
hopper receives a new portion
of the
theproduct,
product,asasdenoted
denoted in Figure
in Figure 2. Once all theallhoppers
2. Once are fullare
the hoppers (weighing and booster),
full (weighing and
the machine calculates the combinations of all the weights, selecting a subset H 0 of k
booster), the machine calculates the combinations of all the weights, selecting a subset
𝐻 of 𝑘 hoppers
hoppers whose sum of the
whose sumweights
of the is greaterisand
weights closerand
greater to the target
closer weight
to the T. weight
target Then, the𝑇.
unloaded
Then, hoppers are
the unloaded filled with
hoppers a newwith
are filled product
a new to repeat
productthe
toprocedure
repeat theuntil completing
procedure until
the Q number
completing theof𝑄packages
number required.
of packages required.

Figure 2. Upright and diagonal multihead weigher system and components.

The w𝑤i weights

The weightsin the hoppers
in the are arranged
hoppers in a normal
are arranged in a distribution wi ∼ N (µ,𝑤σ~𝑁(𝜇,
normal distribution ) [14–16],
𝜎)
where is the average weight of the product to be provided in each weighing
[14–16], where 𝜇 is the average weight of the product to be provided in each weighing
µ hopper H Wi , and
σ is the standard
hopper 𝐻 , anddeviation. Previous deviation.
𝜎 is the standard studies have found that
Previous σ hashave
studies a direct correlation
found that 𝜎 haswitha
the mean quantity and linearly depends on it, due to vibrating feeding systems,
direct correlation with the mean quantity 𝜇 and linearly depends on it, due to vibrating
µ according to
= where is a proportionality coefficient and varies according to the
feeding systems, according to 𝜎 = 𝛾𝜇; where 𝛾 is a proportionality coefficient and varies
σ γµ; γ characteristics of
the packaging
according product
to the [14–17,19].of the packaging product [14–17,19].
characteristics
For each discharge,the
For each discharge, thesingle-layer
single-layermachine
machine performs a number
performs of combinations
a number (NC)
of combinations
determined by Equation (1), where n is the total number of hoppers in the system and k is
(NC) determined by Equation (1), where 𝑛 is the total number of hoppers in the system
the hoppers to be combined. By adding the booster layers, the NCs for the double-layered
and 𝑘 is the hoppers to be combined. By adding the booster layers, the NCs for the
upright and diagonal machines increase, being greater in the latter.
double-layered upright and diagonal machines increase, being greater in the latter.
n𝑛
 
NC 𝑁𝐶= = (1)
k𝑘
For each package produced, the double-layered upright machine can discharge
For each package produced, the double-layered upright machine can discharge booster
booster hoppers or simultaneously discharge the weighing hopper and its corresponding
hoppers or simultaneously discharge the weighing hopper and its corresponding booster
booster hopper (Figure 3), making the NC greater, thus improving the possibility of
hopper (Figure 3), making the NC greater, thus improving the possibility of finding a
finding a subset of hoppers that meet the target weight requirement. Depending on the
subset of hoppers that meet the target weight requirement. Depending on the number
number of weighing hoppers, we have several types of combinations. For example, the
of weighing hoppers, we have several types of combinations. For example, the system
system
can combine k booster𝑘 hoppers
can combine booster hoppers or 𝑖 ≤ hoppers,
or i ≤ 2k booster booster hoppers,
where i iswhere 𝑖 is an
an integer, integer,
with their
with their corresponding weighing hoppers and the other 𝑛 − 𝑖
corresponding weighing hoppers and the other n − i booster hoppers. booster hoppers.
Mathematics 2021, 9,
Mathematics 2021, 9, 1039
x FOR PEER REVIEW 55 of
of 20
21

Figure 3.
Figure 3. Combination of k𝑘== 33for
Combination of forupright
uprightdouble-layered
double-layeredmachines.
machines.
 
For k𝑘== 2,
2,when
when combining
combining the
the booster n combinations are obtained, but
For booster hoppers,
hoppers, combinations are obtained,
the weighing hopper with its corresponding booster can 2 also be combined; therefore,
but the weighing hopper with its corresponding booster can also be combined; therefore,
there are 𝑛 additional cases (see Equation (2)). In the same sense, for 𝑘 = 3, first, there
there are n additional cases (see Equation (2)). In the same sense, for k = 3, first, there are
are  booster hopper combinations. In addition, the weighing hoppers can be combined

n
booster
with booster hopper and
hoppers, combinations. In addition,
each pair can with 𝑛 −hoppers
the weighing
be associated 1 hoppers,
can be combined𝑛
replicated
3
times (see Equation (3)). By mathematical induction of Equations (2)–(8), it is deduced
with booster hoppers, and each pair can be associated with n − 1 hoppers, replicated
that, in general, the total NC of 𝑘 ≤ 𝑛 hoppers is determined by Equation (9),
n times (see Equation (3)). By mathematical induction of Equations (2)–(8), it is deduced
wherein 𝑘general,
that, 2 represents theNC
the total integer
of k part
≤ n of .
hoppers is determined by Equation (9), where
〚k/2〛 represents the integer part of 2k . 𝑛
𝑘 = 2, 𝑁𝐶 = +𝑛 (2)
 2

n
k = 2, NC = 𝑛 𝑛+−n1 (2)
𝑘 = 3, 𝑁𝐶 = 2+ ∙𝑛 (3)
3 1
𝑛 n 𝑛−1 n−1𝑛 𝑛−2
   
𝑘 =k4,= 3, 𝑁𝐶 NC== + + ∙𝑛+ ·∙n (4)
(3)
4 3 2 1 2 0
 𝑛
n−𝑛 1 − 1 ∙ 𝑛 + n 𝑛 ∙ 𝑛n−−22
     
n𝑁𝐶 =
k = 4, 𝑘NC= 5,
= + 5 + · n + · (5)
(4)
4 2 3 22 10
𝑛 𝑛 − 1  𝑛 𝑛−  2 𝑛 𝑛 − 3
𝑘 = 6, 𝑁𝐶 = +
n n∙ 𝑛−+1⋯ + ∙ n +n−2 ∙ (6)
k = 5, NC = 6 4
+ ·n +2 2· 3 0 (5)
5 3 2 1
𝑛 𝑛 − 1 𝑛 𝑛 − 2 𝑛 𝑛 − 3 
𝑘 = 7,  n𝑁𝐶 =  n − +1  ∙ 𝑛 +⋯n+  ∙n − 2  + 
n
∙ 
n − (7)
k = 6, NC = + 7 7−·n 2+ · · · + ·2 7 − 4 + 3 · 7 − 63 (6)
6 4 2 2 3 0
𝑛 𝑛−1 𝑛  𝑛 −2  𝑛 ∙ 𝑛 − 3 𝑛
𝑘 = 8, 𝑁𝐶=n  + n − 1 ∙ 𝑛 + ∙ n +n ⋯
− +  + −
k = 7, NC = 8 + 8 − 2 ·n + · ·2· + 8 − 4· 2 3+ 8 −·6
n n 4
3
(7)
(8)
7 𝑛 7−−42 2 7−4 3 7−6
  ∙ 
n − 18 − 8 n
         
n n−2 n n−3 n n−4
k = 8, NC =
8
+
8−2
·n +
2
·
8−4
+···+
3
·
8−6
+
4
·
8−8
(8)
𝑛 𝑛−𝑖  (9)
𝑁𝐶J /=
2K k  
n n−i
NC = ∑ 𝑖 𝑘 − 2𝑖 (9)
i =0
i k − 2i
Furthermore,for
Furthermore, fordouble-layered
double-layered diagonal
diagonal machines,
machines, only only booster
booster hoppers,
hoppers, only
only weigh-
weighing hoppers, or a combination of both levels can be discharged to produce
ing hoppers, or a combination of both levels can be discharged to produce each package, each
package,
as long asas long as a hopper
a weighing weighing hopper
and and its
its booster arebooster are not simultaneously
not simultaneously unloaded,
unloaded, as denoted
as denoted
in Figure 4. in Figure 4.
For example, for 𝑘 = 2, weighing hopper combinations are generated, each of
the weighing hoppers can be associated with 𝑛 − 1 booster hoppers replicated 𝑛 times
and, finally, booster hopper combinations, the mathematical expression to generate
the combinations is shown in Equation (10). For 𝑘 = 3, only three booster hoppers can be
combined, i.e., two booster hoppers with one of the 𝑛 − 2 remaining weighing hoppers,
 5 1 4 2 3 (13)
𝑛 𝑛−3 𝑛 𝑛−4 𝑛 𝑛−5
…+ ∙ + ∙ + ∙
3 2 4 1 5 0

𝑛 𝑛−𝑖
Mathematics 2021, 9, 1039 𝑁𝐶 = 6(14)
of 20
𝑖 𝑘−𝑖

Dischargeoptions
Figure4.4.Discharge
Figure for𝑘k==33 in
optionsfor in double-layered
double-layered diagonal
diagonal machines.
machines.
 
n
Table 1.For example,
Number for k = 2, per package
of combinations for 𝑛 =hopper
weighing combinations
16 in standard are generated,
(single-layered), double- each
2
layered upright, and double-layered diagonal machines.
of the weighing hoppers
 can be associated with n − 1 booster hoppers replicated n times
n Double-Layered Double-Layered
and, finally,
k booster hopper
Single combinations,
Layered the mathematical expression to generate
2 Upright Diagonal
the combinations
2 is shown in Equation
120 (10). For k = 136
3, only three booster hoppers
480 can be
combined, i.e., two booster hoppers with one of the n − 2 remaining weighing hoppers, one
3 560 800 4480
booster hopper with a combination of two of the n − 1 remaining weighing hoppers, or only
three weighing hoppers. Equations (10)–(13) denote the calculation of the combinations
for k = 2, 3, 4, and 5. In general, by mathematical induction, the total NC of k hoppers with
k ≤ n is determined by Equation (14). Table 1 below presents an example of the NC that
can be generated for each machine type for n = 16 with k = 2 up to k = 8.
      
n n n n
k = 2, NC = + · + (10)
2 1 1 2
         
n n n−1 n n−2 n
k = 3, NC = + · + · + (11)
3 1 2 2 1 3
             
n n n−1 n n−2 n n−3 n n−4
k = 4, NC =
4
+
1
·
3
+
2
·
2
+···+
3
·
1
+
4
·
0
(12)
       
n n n−1 n n−2
k = 5, NC = + · + · +···
  5  1  4  2  3  (13)
n n−3 n n−4 n n−5
...+ · + · + ·
3 2 4 1 5 0

k   
n n−i
NC = ∑ i k−i
(14)
i =0
Mathematics 2021, 9, 1039 7 of 20

Table 1. Number of combinations per package for n = 16 in standard (single-layered), double-layered

upright, and double-layered diagonal machines.
Mathematics 2021, 9, x FOR PEER REVIEW 7 of 21
k Single Layered Double-Layered Upright Double-Layered Diagonal
2 120 136 480
3 4 560 1820 800 3620 29,120
4480
4 5 1820 4368 3620 13,328 29,120
139,776
5 6 4368 8008 13,328 41,328 139,776
512,512
6 7 8008 11,440 41,328 110,448 512,512
1,464,320
7 11,440 110,448 1,464,320
8 12,870 258,570 3,294,720
8 12,870 258,570 3,294,720
9 11,440 536,640 5,857,280
9 11,440 536,640 5,857,280
10 10 8008 8008 996,216 996,216 8,200,192
8,200,192
11 11 4368 4368 1,665,4561,665,456 8,945,664
8,945,664
12 12 1820 1820 2,520,3362,520,336 7,454,720
7,454,720
13 13 560 560 3,465,8403,465,840 4,587,520
4,587,520
14 14 120 120 4,343,1604,343,160 1,966,080
1,966,080
15 15 16 16 4,969,1524,969,152 524,288
524,288
16 16 1 1 5,196,6275,196,627 65,536
65,536

As aforementioned,
As aforementioned, inin
itsits initial
initial configuration,
configuration, the double-layered
the double-layered uprightupright
machinemachine
has
has aa restriction inthe
restriction in thecombination
combination of the
of the k system
𝑘 system hoppers
hoppers (𝐻 ().HSpecifically,
m ). Specifically, a hopper of
a hopper
HofWi𝐻cancanonly be be
only partpart a ka−𝑘nupla
of of − 𝑛𝑢𝑝𝑙𝑎 of hoppers
of hoppers if and
if andonly
onlyififthe
theset thek 𝑘hoppers
setofofthe hopperscontains
H Bj , with𝐻j =, with
contains n + i.𝑗 = 𝑛 + 𝑖. However,
However, in the double-layered
in the double-layered diagonaldiagonal machine,
machine, all HWi 𝐻 HBj that
all and
and 𝐻 to that
belong the kbelong
− nuplato the
must 𝑘− 𝑛𝑢𝑝𝑙𝑎
meet themust meet the
condition of i condition of 𝑖 ≠ 𝑗3−below
6= j − n. Figure 𝑛. Figure 3
illustrates the
below illustrates
invalid combinationsthe invalid
for each combinations
machine. For for example,
each machine.
for n For= 10example,
with k = for3, 𝑛Hopper
= 10 3 in a
with 𝑘 = 3, Hopper
double-layered 3 in system
upright a double-layered
is a weighingupright system
hopper, is a weighing
while Hopper 13 hopper, while hopper.
is a booster
Hopper
Here, 13 is a booster
Hopper 3 may be hopper. Here,
selected forHopper 3 may beprovided
combination selected for combination
that Hopper 13provided
is also selected,
that Hopper 13 is also selected, while, for the diagonal system, they cannot be selected
while, for the diagonal system, they cannot be selected simultaneously. Examples 7 and 8
simultaneously. Examples 7 and 8 of Figure 5 denote invalid combinations for the upright
of Figure 5 denote invalid combinations for the upright machine when trying to combine
machine when trying to combine weighing hoppers without considering their
weighing
correspondinghoppers without
booster considering
hopper. their corresponding
Furthermore, booster hopper.
Example 9 illustrates Furthermore,
an incorrect
Example
combination in the diagonal machine when trying to simultaneously discharge thetrying to
9 illustrates an incorrect combination in the diagonal machine when
simultaneously
weighing hopperdischarge
with their the weighing hopper
corresponding with their corresponding booster hopper.
booster hopper.

Figure Combinations
Figure 5. Combinations notnot allowed
allowed in theindouble-layered
the double-layered
upright upright and double-layered
and double-layered diagonal diagonal
machine configurations.
machine configurations.
Mathematics 2021, 9, 1039 8 of 20

4. Hopper-Filling Strategies
In the industrial field, the filling configuration of the hoppers during the multihead
weighing process is made based on the experience and skill of the machine operator. This
is since the machine does not have a predetermined configuration strategy for all the types
of existing products that can be packaged in it. To address this fact, the present research
proposes solving the packaging problem through an algorithm for different k hopper com-
binations, with an average feed for each hopper according to the three filling strategies
proposed by [14,19,20] for single-layered weighers (see Table 2, Equations (15)–(17)). Two
of these strategies consider the cases in which each hopper i is filled with a different average
quantity of product µi (instead of a common value µ = T/k). In this sense, and considering
the original concept of each filling strategy, this study will explore the cases in which
hopper feeds are defined so that hopper subgroups may share the same value of µi . These
strategies have been implemented in single-layered machines using single-target and bio-
target algorithms with good results in reducing process variability [14,18–20]. However,
they have not yet been tested on double-layered machines. With the implementation of the
strategies, it is also intended to evaluate if it is more efficient for the process to supply all
hoppers with the same amount of product or uneven amount of product according to the
coefficient of proportionality γ related to the product to be packed. Hopper feed variation
will depend on a change value represented by the δ parameter, which is commonly used
in statistical process control to simulate out-of-control processes. Nevertheless, for the
purposes hereof, it will be used to voluntarily simulate an uneven product supply for
weighing hoppers. Herein, δ will take values from 0 to 3 with increments of 0.5 (these in-
crements are represented by the δmin parameter, as expressed in Equation (15)). In addition,
the average filling value for each hopper will also be influenced by the proportionality
coefficient γ (σ = γµ), which will depend on the type of material to be packaged. In this
document, proportionality coefficients are used for two different pasta products: Fusilli,
with γ = 0.123, and Ravioli, with γ = 0.331, [19,20]. For example, to calculate the standard
deviation for a target weight of 500 g when we want to pack a product such as Fusilli by
500 g
combining three hoppers, we will have a theoretical deviation of σ = 0.123· 3 = 20.5 g.

Table 2. Number of hoppers for each group according to the S1 , S2 , and S3 strategies and distribution of hoppers.

Distribution Strategy n1 n2 n3 n4 n5 a=mod( n5 )

n/5 n/5 n/5 n/5 n/5 a =0
〚n/5〛 〚n/5〛 〚n/5〛+ a 〚n/5〛 〚n/5〛 a =1
S1 〚n/5〛+ 1 〚n/5〛 〚n/5〛 〚n/5〛 〚n/5〛+ 1 a =2
Equal 〚n/5〛+ 1 〚n/5〛 〚n/5〛+ 1 〚n/5〛 〚n/5〛+ 1 a =3
〚n/5〛+ 1 〚n/5〛+ 1 〚n/5〛 〚n/5〛+ 1 〚n/5〛+ 1 a =4
S2 〚n/3〛 n − 2·〚n/3〛 〚n/3〛
S3 n
S1 1 1 n−4 1 1
Central 1, i f n ≤ 8 1, i f n ≤ 8
S2 n − 2n1
2, i f n > 8 2, i f n > 8
S3 n
S1 (n − 2)/2 1 0 1 (n − 2)/2
Extreme S2 (n − 2)/2 2 (n − 2)/2
S3 n

In this way, to assess the algorithm proposed for double-layered machines, the three
strategies will determine the average product supply for hopper subgroups and the to-
tal number of hoppers in each subgroup. The first strategy (S1 ) proposes dividing the
n weighing hopper into five groups, (n1 , n2 , n3 , n4 , n5 , where ∑5i=1 ni = n) and feeding
different average quantities to each subgroup (µ1 , µ2 , µ3 , µ4 , µ5 ), establishing the filling
configuration during the packaging process according to Equation (15). The second strategy
(S2 ) proposes diving the n weighing hopper into three groups, (n1 , n3 , n5 ) and feeding
Mathematics 2021, 9, 1039 9 of 20

different average quantities to each subgroup (µ1 , µ3 , µ5 ), as per Equation (16). Finally, in
the third strategy (S3 ), the n weighing hoppers are filled with the same amount of product,
µ = Tk for each hopper, as expressed in Equation (17).

µ1 = µ − δσ
 

 



 µ2 = µ − (δ − δmin )σ 


µ3 = µ = Tk


wi ∼ N µ j , σ = γµ j = (15)
µ4 = µ + (δ − δmin )σ

 


 

 
µ5 = µ + δσ
 
µ1 = µ − δσ
 
µ3 = µ = Tk


wi ∼ N µ j , σ = γµ j = (16)
µ5 = µ + δσ
 
 
T
wi ∼ N (µ, σ = γµ) = µ = (17)
k
In addition to the S1 , S2 , and S3 filling strategies, in which an average product supply
is established for each subgroup, we must define a distribution of hoppers, in particular,
the number of hoppers that are assigned to each subgroup (see Table 2). In this sense, three
types of distributions are proposed: equal, central, and extreme. In the equal distribution
strategy, each group includes approximately the same number of hoppers. For example,
for
the S1 filling strategy, the total number of hoppers (n) is divided among the five groups
n
5 . Each of the ni (i = 1, . . . 5) is assigned a number of hoppers equal to the largest
integer multiplied by 5 that is closest to n (integer part 〚 n2 〛). If the remainder of the division
mod n5 is 1, the central group (n3 ) will have one




more hopper. If it is 2, a hopper
will be assigned to each
extreme group. For mod n5 = 3, they are distributed between
n1 , n3 , and n5 . If mod n5 = 4, the n3 group will have one less hopper than the rest.
The central distribution consists of assigning as many hoppers as possible to the
central set of hoppers. For example, in strategy S2 for n ≤ 8 and n > 8, one and two
hoppers will be assigned, respectively, to each end, and the surplus is assigned to the
central group. Finally, the extreme distribution assigns the largest number of hoppers to
the extreme subgroups n1 and n5 , and the least amount to n3 . Tables 3 and 4 illustrate
an example of the number of hoppers for each subgroup according to the S1 , S2 , and S3
strategies with n = 16 weighing hoppers, and the average supply values for central, equal,
and extreme subgroup distributions.

Table 3. Number of hoppers for each group according to the S1 , S2 , and S3 strategies for equal,
central, and extreme distributions with n = 16.

Distribution Strategy n1 n2 n3 n4 n5
S1 3 3 4 3 3
Equal S2 5 6 5
S3 16
S1 1 1 12 1 1
Central S2 2 12 2
S3 16
S1 7 1 0 1 7
Extreme S2 7 2 7
S3 16
Mathematics 2021, 9, 1039 10 of 20

Table 4. Average filling weight according to the S1 , S2 , and S3 strategies for equal, central, and extreme distributions
with n = 16.

Distribution Strategy Average Filling Weight

S1 µ1 µ1 µ1 µ2 µ2 µ2 µ3 µ3 µ3 µ3 µ4 µ4 µ4 µ5 µ5 µ5
Equal S2 µ1 µ1 µ1 µ1 µ1 µ3 µ3 µ3 µ3 µ3 µ3 µ5 µ5 µ5 µ5 µ5
S3 µ3 µ3 µ3 µ3 µ3 µ3 µ3 µ3 µ3 µ3 µ3 µ3 µ3 µ3 µ3 µ3
S1 µ1 µ2 µ3 µ3 µ3 µ3 µ3 µ3 µ3 µ3 µ3 µ3 µ3 µ3 µ4 µ5
Central S2 µ1 µ1 µ3 µ3 µ3 µ3 µ3 µ3 µ3 µ3 µ3 µ3 µ3 µ3 µ5 µ5
S3 µ3 µ3 µ3 µ3 µ3 µ3 µ3 µ3 µ3 µ3 µ3 µ3 µ3 µ3 µ3 µ3
S1 µ1 µ1 µ1 µ1 µ1 µ1 µ1 µ2 µ4 µ5 µ5 µ5 µ5 µ5 µ5 µ5
Extreme S2 µ1 µ1 µ1 µ1 µ1 µ1 µ1 µ3 µ3 µ5 µ5 µ5 µ5 µ5 µ5 µ5
S3 µ3 µ3 µ3 µ3 µ3 µ3 µ3 µ3 µ3 µ3 µ3 µ3 µ3 µ3 µ3 µ3
Weighing hoppers: HW 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Booster hoppers: HB 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

5. Optimization Model and Packaging Algorithm

This section presents the mathematical model and the packaging algorithm for the
process in double-layered upright and diagonal machines. In this work, an adaptation of
the well-known backpack problem is used. For each package, a deterministic knapsack
problem is solved. Some researchers have optimized this way of determining the final
weight of the package [21–23].
The symbology, decision variables, target function, and constraints for each process
are also presented.

5.1. Symbology
Q : number of packages;
` : iteration in which the package is packed, where ` ∈= {1, 2, 3.Q};
HWi : set of n weighing hoppers, i = {1, 2, 3.n};
HBj : set of n booster hoppers, j = {n + 1, n + 2, . . . 2n};

Hm = HWi ∪ HBj : set of the 2n hoppers m = {1, 2, 3, . . . , 2n};
H 0 : subset of combined hoppers;
T : target label weight;
k : number of hoppers to combine;
wi : actual weight of each hopper i = {1, 2, 3, . . . 2n} based on filling strategy (S1 , S2 , or S3 );
W` : sum of the weights of the k hoppers in the ` iteration.

5.2. Target Function

The target function for the packaging processes in double-layered machines seeks to
minimize the difference between the effective content of the package and its target weight
T. The binary vectors are defined for the weighing hoppers (Equation (18)) and Yj for
the booster hoppers (Equation (19)), whose components xi or y j take the value of 1 if the
hopper weight Hm was selected, or else it takes the value of 0 (Equations (20) and (21)).

Xi = ( x 1 , x 2 , x 3 , x 4 , . . . x n ) (18)

Yj = (yn+1 , yn+2 , yn+3 , yn+4 . . . y2n ) (19)


1, if HWi is the selected hopper
xi = (20)
0, otherwise

1, if HBi is the selected hopper
yi = (21)
0, otherwise
Mathematics 2021, 9, 1039 11 of 20

To select the hoppers, whose sum of the weights are the closest and greatest to the
package weight, the function will be minimized, f = W` − T ≥ 0.
" #
n 2n
minimize f ( x, y) = ∑ x i wi + ∑ yj wj − T (22)
i =1 j = n +1

The problem restrictions will be directly determined by hopper selection restrictions.

For the double-layered upright machine, the operating restriction of the system is that an
upper hopper cannot be selected if its corresponding booster hopper has not been selected,
which means that Equation (23) must be used. For the double-layered diagonal machine, a
weighing hopper and its booster cannot be selected simultaneously, which is guaranteed
by Equation (24).
xi − yi ≤ 0 (23)
xi + yi ≤ 1 (24)
In each ` iteration, the best combination of k hoppers from the 2n available must be
selected. If Equation (22) and the restrictions Equations (20), (21), and (23) for the double-
layered upright machine are satisfied simultaneously for the whole set of k combined
hoppers and Equations (20), (21), and (24) for the double-layered diagonal machine, then
the combination is valid.
As an example of the description of the mathematical model adjusted to the S2 , first,
the vectors are defined for each subgroup (Equations (25)–(30)) according to Equations
(31)–(36). The mathematical model is expressed as a function to minimize (Equation (37)),
subject to the restrictions from Equations (38)–(40) for the double-layered upright machine
and from Equations (41)–(43) for the double-layered diagonal.
n
S 1
 
Xi 2 = xiS2 , . . . , xnS12 (25)

n
S 2
 
Xm2 = xnS2 +1 , . . . , xnS12 +n2 (26)
1
n
S 3
 
Xr 2 = xnS2 +n2 +1 , . . . , xnS2 (27)
1
n
S 1
 
Yj 2 = y1S+
2
n , . . . , y S2
n + n1 (28)
n
S 2
 
Yt 2 = ySn2+n S2
+1 , . . . , y n + n1 + n2 (29)
1
n
S 3
 
Yu 2 = ySn2+n +n2 +1 , . . . , y2n
S2
(30)
1
( n
S2 1
xi = 1, if H Wi is chosen (31)
0, in other case
( n
S2 2
xm = 1, if H Wm is chosen (32)
0, in other case
( n
S2 3
xr = 1, if HWr is selected (33)
0, in other case
n
S2 1
(
yj = 1, if H Bj is selected (34)
0, in other case
( n
S2 2
yt = 1, if H Bt is chosen (35)
0, in other case
Mathematics 2021, 9, 1039 12 of 20

( n3
S
yu = 1, if HBu2 is selected (36)
0, in other case
" #
n1 n1 + n2 n n1 + n n1 + n2 + n 2n
∑ xl1 w1l + ∑ 2 2
xm wm + ∑ xr3 wr3 + ∑ y1j w1j + ∑ y2t w2t + ∑ y3u w3u −T ≥0 (37)
l =1 m = n1 +1 r = n1 + n2 +1 j=l +n t = n1 + n u = n1 + n2 + n

xi − y j ≤ 0 (38)
xm − yt ≤ 0 (39)
xr − yr ≤ 0 (40)
xi + y j ≤ 1 (41)
xm + yt ≤ 1 (42)
xr + yr ≤ 1 (43)

5.3. Algorithm
In this subsection, we present the algorithm designed for the packaging process for
double-layered upright and diagonal machines. The step-by-step process is generic for any
of the k combinations of the 2n hoppers of the system, with initial constants adjustable to
the type of product packaged.
• Step 1. The initial values and conditions are defined.
Q : Total number of packages processed;
n : Number of weighing hoppers;
k : Number of hoppers to combine 2 ≤ k ≤ n;
T : Target weight > 0.
• Step 2. The empty weighing hoppers HWi are loaded with randomly assigned weights
wi , i ∈ {1, 2, . . . n} according to S1, S2, or S3 and according to the number of hoppers
in each group.
• Step 3. The contents of the weighing hoppers HWi are discharged into their corre-
sponding empty booster hoppers HBj , with wn+i = wi and wi = 0.
• Step 4. The weighing hoppers previously discharged are reloaded.
• Step 5. The hoppers that meet the criteria from Equation (23) or Equation (24) are
combined for the upright or diagonal machines, respectively.
• Step 6. The difference between the sum of the weights of each combination and the
target weight is calculated (Equation (22)).
• Step 7. The subset H 0 of k hoppers whose difference with T is minimal, from those
calculated in the previous point, is selected.
• Step 8. The product is discharged and packed.
• Step 9. If the required number of Q packets has been completed, the process ends.
Otherwise, it returns to Step 2.

6. Preliminary Analysis
To validate the algorithm, we used proportionality coefficients of γ = 0.123 for Fusilli
and γ = 0.331 for Ravioli [20]. In addition, a number of n = 16 weighing hoppers were
assessed, at k = 2, 3, 4, 5, and 7, δ = 2, δmin = 0.5, and a target weight of T = 250 g.
The number of hoppers in each subgroup is shown in Table 3. Combinations generated
for the different values of k assessed are presented in Table 1. The performance measures
calculated were the average weight of the packages produced (µ paq ), the standard devi-
ation of the
 packages produced
 (σpaq ), and the coefficient of variation of the packages
paq µ
produced CVpaq = σpaq , all above for Q = 10,000 packages.
Results for the double-layered upright and diagonal machine are presented in
Tables 5 and 6. Here, we observe that at γ = 0.123, for the double-layered upright machine,
the S3 strategy produces a weight closer to T and lower values of σpaq and CVpaq , when
Mathematics 2021, 9, 1039 13 of 20

k = 7. However, this strategy offers the highest values for µ paq and σpaq , thus becoming
the least favorable for the process. In S1 and S2 , the average weight closest to T and lower
values of σpaq and CVpaq were obtained at k = 7. However, their values were similar to k = 6.
These last strategies seem to be the most convenient in terms of reducing process variability.
This behavior of the strategies is maintained for the double-layered diagonal machine,
which also denotes lower values when compared to those obtained in the double-layered
upright machine.

Table 5. Results of S1 , S2 , and S3 strategies, for k = 2, 3, 4, 5, 6, and 7, and γ = 0.123.

FUSILLI
Vertical Diagonal
k µpaq σpaq CVpaq µpaq σpaq CVpaq µ1 µ2 µ3 µ4 µ5
2 254.158 2.855 0.0110 253.650 2.869 0.0110 94.25 101.94 125.00 148.06 155.75
3 250.231 0.208 0.0008 250.063 0.049 0.0002 62.83 67.96 83.33 98.71 103.83
4 250.031 0.029 0.0001 250.005 0.004 1.62e-05 47.13 50.97 62.50 74.03 77.88
S1
5 250.006 0.006 2.36e-05 250.001 0.001 2.48e-06 37.70 40.78 50.00 59.23 62.30
6 250.002 0.002 6.51e-06 250.000 1.36e-04 5.47e-07 31.42 33.98 41.67 49.35 51.92
7 250.001 0.001 2.42e-06 250.000 5.10e-05 2.04e-07 26.93 29.13 35.71 42.30 44.50
2 253.928 2.82 0.0111 253.319 2.691 0.001 94.25 125.00 155.75
3 252.181 2.158 0.0086 250.056 0.044 0.0002 62.83 83.33 103.83
4 250.031 0.028 0.0001 250.004 0.004 1.34e-05 47.13 62.50 77.88
S2
5 250.006 0.006 2.42e-05 250.001 0.001 2.02e-06 37.70 50.00 62.30
6 250.002 0.002 6.68e-06 250.001 1.21e-04 4.85e-07 31.42 41.67 51.92
7 250.001 0.001 2.04e-06 250.000 4.21e-05 1.68e-07 26.93 35.71 44.50
2 253.405 2.662 0.0105 252.615 2.392 0.0095 125.00
3 250.414 0.82 0.0033 250.170 0.673 0.0027 83.33
4 250.161 0.657 0.0026 250.070 0.482 0.0019 62.50
S3 5 250.138 0.725 0.0029 250.107 0.720 0.0029 50.00
6 250.166 0.885 0.0040 250.060 0.556 0.0022 41.67
7 250.084 0.623 0.0024 250.071 0.609 0.0024 35.714

Table 6. Results of the S1 , S2 , and S3 strategies, for k = 3, 4, 5, 6, and 7, and γ = 0.331.

RAVIOLI
Vertical Diagonal
k µpaq σpaq CVpaq µpaq σpaq CVpaq µ1 µ2 µ3 µ4 µ5
2 254.527 2.929 0.0120 254.138 2.854 0.0110 42.25 62.94 125.00 187.06 207.75
3 252.231 2.215 0.0088 251.915 2.199 0.0087 28.17 41.96 83.33 124.71 138.50
4 250.102 0.104 0.0004 250.028 0.038 0.0002 21.13 31.47 62.50 93.53 103.88
S1
5 250.018 0.017 0.0001 250.002 0.002 7.47e-06 16.90 25.18 50.00 74.83 83.10
6 250.005 0.005 1.85e-05 250.000 4.29e-04 1.69e-06 14.08 20.98 41.67 62.35 69.25
7 250.002 0.002 6.86e-06 250.000 1.12e-04 4.84e-07 12.07 17.98 35.71 53.45 59.36
2 254.587 2.867 0.0113 254.154 2.857 0.0112 42.25 125.00 207.75
3 252.539 2.421 0.0096 251.638 2.086 0.0083 28.17 83.33 138.50
4 250.113 0.118 0.0005 250.017 0.016 0.0001 21.13 62.50 103.88
S2
5 250.018 0.017 0.0001 250.002 0.002 6.85e-06 16.90 50.00 83.10
6 250.005 0.004 1.72e-05 250.000 3.39e-04 1.35e-06 14.08 41.67 69.25
7 250.001 0.001 5.95e-06 250.000 1.22e-04 4.88e-07 12.07 35.71 59.36
2 254.422 2.867 0.0113 253.58 2.715 0.0107 125.00
3 250.887 1.376 0.0055 250.245 0.597 0.0024 83.33
4 250.318 0.960 0.0038 250.102 0.501 0.0020 62.50
S3 5 250.170 0.691 0.0028 250.053 0.410 0.0016 50.00
6 250.160 0.793 0.0030 250.049 0.443 0.0017 41.667
7 250.069 0.505 0.0020 250.064 0.565 0.0022 35.714

However, the results of the tests for γ = 0.331 show that the average weight of
the packages closest to the target weight and the lowest values of σpaq and CVpaq are
presented for k = 6 and 7 in both machines, being lower in S1 and S2 . In general terms, we
observe that when the value of the proportionality coefficient is increased to γ = 0.331, the
µ paq , σpaq , and CVpaq values also increase. However, the behavior (operation) of the process
is similar for both products. Based on these results, the design of experiments presented in
Mathematics 2021, 9, 1039 14 of 20

Section 7 was proposed to determine which conditions decrease process variability and
provide higher values and as close as possible to the target weight.

7. Experiment Design
To follow up on our preliminary analysis, an experiment design (DOE) of fixed effects
factors [24] was conducted to determine the best combination of treatments that provides
the least process variability. For the design, the factor levels that evidenced the best
performance in their coefficient of variation are considered, discarding those that yielded
the lowest levels of process variability reduction. The multifactorial design consists of
eight factors, as shown in Table 7, for a total of 1512 treatments in each machine, each of
them with three replicates, finally obtaining 12,096 runs. The response variable studied is
the coefficient of variation obtained in each run of 10,000 packages since it is a measure
that may be used to compare different target weights T. In addition to analyzing the best
factor combination, the upright and diagonal machines are compared against each other
to determine which offers less process variability. Here, the first factor encompasses two
levels, which refer to the double-layered upright or diagonal machine. A second factor
is associated with the number of weighing hoppers. In this case, machines with 10, 14,
and 16 weighing hoppers were assessed. The third factor is the number of hoppers to
combine with values from k = 5 to k = 7. Another factor to consider is the target label
weight, which is set at two levels: 250 g and 500 g. In addition, two packaging products,
Fusilli and Ravioli, constitute the two levels of the sixth factor. The S1 and S2 strategies
determine hopper subgroups, grouped by equal, central, and extreme levels. The filling
position constitutes the eighth factor with seven levels δmin = 0, 0.5, 1, 1.5, 2, 2.5, and 3,
which simulate an out-of-control process when δmin > 0.

Table 7. Experiment design factors and levels.

Factors Factor Levels

Machine type Upright Diagonal
Number of weighing hoppers: n 10 14 16
Number of hoppers to combine k 5 6 7
Target weight: T 250 500
Material type: gamma (γ) Fusilli = 0.123 Ravioli = 0.331
Strategy: Number of subgroups 3 5
Hopper distribution Central Equal Extreme
Filling position: Delta (δ) 0 0.5 1 1.5 2 2.5 3

Given that the data of the response variation coefficient variable (CVpaq) are asymmet-
ric, with a lower target weight, adjusted box plots are used to compare the levels within the
same factor, as shown in Figure 6. To assess the experiment design, a transformation of the
variable CV paq by Johnson’s method [25] is used to standardize the data, thus obtaining
the transformed coefficient of variation (CV paqT) as the response variable.
Results from the experiment design were analyzed using analysis of variance (ANOVA).
Table 8 denotes the results of the ANOVA. As it may be observed, no significant differences
were found in the product type and number of subgroups factors for the hopper distribu-
tion. Figure 7 denotes the significant factors. When analyzing the interactions, significant
p-values were found for Machine Type Target Weight, Machine Type· k, Target Weight·n, n·k, n
k, n·Delta, n·Gamma, n·Hopper Distribution, k·Delta, k·Number of Subgroups, k·Hopper Distribu-
tion, Delta·Gamma, Delta·Number of Subgroups, and Number of Subgroups·Hopper Distribution.
Based on the results of the significant interactions between the design factors
(Figures 8 and 9), we can infer that when comparing machine performances according to the
number of weighing hoppers, n = 16 provides greater accuracy, and it is even better when
the target weight is 500 g. For the number of hoppers combined, significant differences
were found between the mean values of the coefficient of variation. Figures 6 and 7 denote
that for almost all interactions, k = 7 is the ideal number of hoppers to reach the target
weight, except at δ = 1.5, which suggests the use of six hopper combinations. Significant
differences were also found for CV paqT, according to the target weight, exhibiting better
Mathematics 2021, 9, 1039 15 of 20

behaviors
Mathematics 2021, 9, x FOR PEER REVIEW if T = 500 g. Regarding the S1 and S2 filling strategies, five subgroups 15 provided
of 21
better results at δ = 2.5. In addition, significant differences were found regarding the
product type. In this aspect, better results were found when the proportionality coefficient
is the lowest because
Given that the lower
data ofvalues of γ introduce
the response less variability
variation coefficient in the weights
variable supplied
(CVpaq) are
toasymmetric,
the weighing hoppers. Here, Fusilli (γ = 0.123) reported less process variability
with a lower target weight, adjusted box plots are used to compare the levels with
anwithin
optimum
the same at δ =as2.5.
point factor, Regarding
shown in Figurethe 6.
hopper distribution
To assess for eachdesign,
the experiment group, athe
central strategy denotes
transformation better 𝐶𝑉𝑝𝑎𝑞
of the variable behaviorby for both machines.
Johnson’s Hence,
method [25] we
is used to can conclude
standardize thethat
the optimal
data, process uses
thus obtaining the the diagonal coefficient
transformed = 7, n =(𝐶𝑉𝑝𝑎𝑞𝑇)
machine atofk variation 16 T = 500, S1 =
as the 5, central
response
distribution,
variable. δ = 2.5 (δ = 1.0 for Ravioli), and γ = 0.123.

Figure6.6.Adjusted
Figure Adjustedbox
boxplot
plot of
of the
the coefficient
coefficient of
of variability
variabilityby
byfactors.
factors.

Results
Table from the
8. ANOVA experiment
for CVpaqT - sumdesign were
of squares typeanalyzed
III. using analysis of variance
(ANOVA). Table 8 denotes the results of the ANOVA. As it may be observed, no
Source Sum ofwere
significant differences Squares
found in the Dfproduct Meantype Square of subgroupsp-Value
and numberF-Ratio factors
MAIN EFFECTS for the hopper distribution. Figure 7 denotes the significant factors. When analyzing the
Machine type 1.72e14
interactions, significant p-values were found 1 1.72e14
for Machine Type Target 327.78
Weight, Machine 0.0000
Type·
Target weight 2.28e14 1 2.28e14
k, Target Weight·n, n·k, n k, n·Delta, n·Gamma, n·Hopper Distribution,434.55 0.0000
k·Delta, k·Number of
n 5.14e14 2 2.57e14 488.73 0.0000
k Subgroups, k·Hopper Distribution,
1.90e14 Delta·Gamma,
2 Delta·Number
9.52e13 of Subgroups,
181.24 and Number
0.0000 of
Delta Subgroups·Hopper Distribution.
2.35e14 6 3.92e13 74.68 0.0000
Gamma 4.10e11
Based on the results of the significant 1 interactions 4.10e11 0.78 factors (Figures
between the design 0.3771
Number subgroup strategy 6.17e11 1 6.17e11 1.17 0.2785
Distributions 8 and 9), we can infer that when comparing
1.25e14 2 machine
6.26e13performances
119.06according0.0000
to the
INTERACTIONS number of weighing hoppers, 𝑛 = 16 provides greater accuracy, and it is even better
Machine type target weight
when the target weight 2.83e13is 500 g. For 1the number2.83e13 53.82
of hoppers combined, 0.0000
significant
Machine Type·k 3.95e14 2 1.98e14 376.23 0.0000
Target weight·n
differences were found between the mean
4.20e12 2
values of 2.10e12
the coefficient of4.00
variation. Figures
0.0184
6
Target weight·k and 7 denote that for1.71e13
almost all interactions,
2 𝑘 = 78.54e12
is the ideal number
16.26 of hoppers0.0000 to
n·k reach the target weight,8.94e13except at 𝛿 4= 1.5, which suggests the42.53
2.23e13 use of six 0.0000
hopper
n·Delta 3.62e14
combinations. Significant differences were12 also found 3.02e13 57.44
for 𝐶𝑉𝑝𝑎𝑞𝑇, according 0.0000
to the target
n·Gamma 6.04e12 2 3.02e12 5.74 0.0032
n·Distributions weight, exhibiting better
1.30e13 behaviors if 4 𝑇 = 500 g . Regarding
3.26e12 the 𝑆
6.20 and 𝑆 filling
0.0001
strategies, five subgroups provided better results at 𝛿 = 2.5 . In addition, significant
differences were found regarding the product type. In this aspect, better results were
found when the proportionality coefficient is the lowest because lower values of 𝛾
introduce less variability in the weights supplied to the weighing hoppers. Here, Fusilli
Mathematics 2021, 9, 1039 16 of 20

Table 8. Cont.

Source Sum of Squares Df Mean Square F-Ratio p-Value

k·Delta 2.69e14 12 2.24e13 42.58 0.0000
k·Number subgroup strategy 8.68e12 2 4.34e12 8.26 0.0003
k·distributions 1.46e13 4 3.65e12 6.95 0.0000
Delta·gamma 1.42e14 6 2.37e13 45.09 0.0000
Delta·number subgroup strategy 5.05e13 6 8.41e12 16.01 0.0000
Number subgroup strategy·Distributions 5.04e12 2 2.52e12 4.80 0.0083
Mathematics
Mathematics2021,
2021,9,9,xxFOR
Residual FORPEER
PEERREVIEW
REVIEW 6.32e15 12,018 5.25e11 18
18of
of23
23
Total (corrected)
Means and 95,0 Percent LSD Intervals
9.19e15 12,095
Means and 95,0 Percent LSD Intervals Means
Meansand
and95,0
95,0Percent
PercentLSD
LSDIntervals
Intervals

2323 3131

1818
2121

CVpaqT (X 10,000)
10,000)

CVpaqT (X 10,000)
CVpaqT(X10,000)

1313

88 1111
CVpaqT(X

33
11
-2-2

-7-7 -9-9
Diagonal
Diagonal Upright
Upright 250
250 500
500
Means Machine
Machine Type
Type Means
Meansand Target
and95,0 Weight
Percent LSD
LSDIntervals
Meansand
and95,0
95,0Percent
PercentLSD
LSDIntervals
Intervals Target
95,0 Weight
Percent Intervals

4343 2828

3333
1818
CVpaqT (X 10,000)
10,000)

CVpaqT (X 10,000)
(X10,000)

2323

88
CVpaqT(X

1313
CVpaqT

33
-2-2
-7-7

-17
-17 -12
-12
1010 1414 1616 55 66 77
nn kk
Means
Meansand
and95,0
95,0Percent
PercentLSD
LSDIntervals Means
(X(X10000,0)
10000,0)
Intervals Meansand
and95,0
95,0Percent
PercentLSD
LSDIntervals
Intervals

3434 2525

2424 2020
(X10,000)

CVpaqT(X 10,000)
CVpaqT(X10,000)

CVpaqT(X 10,000)

1515
1414
1010
CVpaqT

44
55

-6-6
00

-16
-16 -5-5
00 0.5
0.5 11 1.5
1.5 22 2.5
2.5 33 Centre
Centre Equal
Equal Extreme
Extreme
Delta
Delta Distributions
Distributions

Figure 7. Means and least significant difference (LSD) intervals for the main significant factors.

Figure
Figure7.7.Means
Meansand
andleast
leastsignificant
significantdifference
difference(LSD)
(LSD)intervals
intervalsfor
forthe
themain
mainsignificant
significantfactors.
factors.
Mathematics 2021, 9, 1039 17 of 20
Mathematics 2021, 9, x FOR PEER REVIEW 19 of 23
Mathematics 2021, 9, x FOR PEER REVIEW 19 of 23

Interactions and 95,0 Percent LSD Intervals Interactions and 95,0 Percent LSD Intervals
Interactions and 95,0 Percent LSD Intervals Interactions and 95,0 Percent LSD Intervals

48 54
48 54
Target Weight Machine Type
Target Weight Machine Type
250 Diagonal
250 Diagonal
500 Upright
28 500 34 Upright

10,000)
34
10,000)

28

CVpaqT(X10,000)
CVpaqT(X10,000)

8 14

CVpaqT(X
CVpaqT(X

8 14

-12 -6
-12 -6

-32 -26
-32 -26
5 6 7 5 6 7
5 6 7 5 6 7
k k
Interactions and 95,0k Percent LSD Intervals k
Interactions and 95,0 Percent LSD Intervals Interactions and 95,0 Percent LSD Intervals
Interactions and 95,0 Percent LSD Intervals

7 6
7 6
n Distributions
n Distributions
10 Centre
5 10 Centre
5 14 Equal
14 4 Equal
16
10,000)

Extreme

10,000)
16 4
CVpaqT(X10,000)

Extreme

CVpaqT(X10,000)
3
3
2
CVpaqT(X

CVpaqT(X 2
1
1

0
-1 0
-1

-3 -2
-3 -2
5 6 7 5 6 7
5 6 7 5 6 7
k k
k
Interactions and 95,0 Percent LSD Intervals Interactions and 95,0 Percent
k LSD Intervals
Interactions and 95,0 Percent LSD Intervals Interactions and 95,0 Percent LSD Intervals

26 43
26 43
Number subgroup Target Weight
Number subgroup Target Weight
3 33 250
3 33 250
5 500
16 5 500
10,000)
10,000)

16
CVpaqT(X10,000)
CVpaqT(X10,000)

23
23

6 13
CVpaqT(X
CVpaqT(X

6 13

3
3
-4
-4
-7
-7

-14 -17
-14 -17
5 6 7 Diagonal Upright
5 6 7 Diagonal Upright
k Machine Type
k Machine Type
Mathematics 2021, 9, x FOR PEER REVIEW 20 of 23

(X 10000,0)
Figure 8. Significant interactions between factors for the CVpaqT (II) variable.
Interactions and 95,0 Percent LSD Intervals Interactions and 95,0 Percent LSD Intervals

42 71
Figure Gamma
Figure 8.
8. Significant
Significant interactions
interactions
0.123
between
between factors
factors for
for the
the CVpaqT
CVpaqT (II)
(II) variable.
variable. Target Weight
32 250
0.331 51 500
CVpaqT (X10,000)

CVpaqT(X 10,000)

22
31
12
11
2

-8 -9

-18 -29
Interactions and 95,014Percent LSD Intervals
10 16 (X 100000,) 10 14 16
Interactions and 95,0 Percent LSD Intervals
n n

57
Figure 9. Cont.8
Distributions k
Centre 5
Equal 6 6
37 Extreme
CVpaqT(X 10,000)

7
CVpaqT (X10,000)

17 2

0
-3
-2
 CVpaqT (X10,

CVpaqT(X 10,
31
12
11
2

-9
Mathematics 2021,-89, 1039 18 of 20
-18 -29
Interactions and 95,014Percent LSD Intervals
10 16 (X 100000,) 10 14 16
Interactions and 95,0 Percent LSD Intervals
n n

57
8
Distributions k
Centre 5
Equal 6 6
37 Extreme
CVpaqT(X 10,000)

CVpaqT (X10,000)
4

17 2

0
-3
-2

-23 -4
10 14 16 0 0.5 1 1.5 2 2.5 3
(X 10000,0) n (X 100000,) Interactions and 95,0Delta
Percent LSD Intervals
Interactions and 95,0 Percent LSD Intervals

49 10
Gamma n
0.123 10
0.331 7 14
29 16
CVpaqT (X10,000)
CVpaqT (X10,000)

4
9
1

-11
-2

-31 -5
0 0.5 1 1.5 2 2.5 3 0 0.5 1 1.5 2 2.5 3
Delta
(X 10000,0) Interactions and 95,0 Percent LSD Intervals (X 10000,0) Interactions and 95,0Delta
Percent LSD Intervals

58 33
Number subgrup Number subgroup
3 3
5 5
38 23
CVpaqT(X 10,000)
CVpaqT (X10,000)

18 13

3
-2

-7
-22
Centre Equal Extreme
0 0.5 1 1.5 2 2.5 3
Distributions
Delta

Figure 9. Significant interactions between factors for the CVpaqT (II) variable.

Figure 9. Significant interactions between factors for the CVpaqT (II) variable.
8. Discussion
Multihead machines for their proper operation require two processes: packaging and
machine configuration. The packaging process is linked to the problem of the backpack,
that is, how to select the hoppers whose sum of the weights is greater and closest possible to
the target weight. The new software, implemented for upright and diagonal double-layer
machines with real weights according to the preliminary results (Tables 5 and 6), shows
that it is capable of responding to the problem by finding the optimal value in each package,
complying with the requirements. On the other hand, the configuration of the machine
is related to the amount of product that each weighing hopper receives, and in this sense,
the filling strategies focus on evaluating the performance of the machine according to the
amount of product supplied to each hopper. The S3 strategy provides each hopper with
the same amount of product, while the S2 divides the hoppers into three groups and in
the same way, S1 into five groups, which guarantees an unequal supply of product to each
subset of hoppers. The preliminary results of Tables 5 and 6 show that the configuration of
Mathematics 2021, 9, 1039 19 of 20

the machines with the lowest performance in terms of the target weight and the variation
of the process is the S3 strategy, which supplies the same quantity of product to all the
hoppers; therefore, it is not taken into account for the design of the experiment (Table 7).
Likewise, in the configuration process, distribution strategies (central, equal, and extreme)
were tested with the number of hoppers for each subset determined in each strategy S1
and S2, according to the group to which the largest number of hoppers is assigned. In the
results of the design of experiments (Table 8), it is observed that there is no difference
between the strategies S1 and S2 (p-value = 0.27); however, when contrasting them in the
presence of other factors, for example, the number of hoppers to combine (k) and the delta
value (δ), we find that in the presence of these factors, there is a significant difference
between them. In the same sense, for the type of product, significant p-values are found
in the presence of a second factor such as the case of delta and the number of hoppers in
the system. In accordance with the above, the type of product is an important factor to
take into account when configuring the machine. When analyzing the interactions of the
other factors with the delta value (δ) (Table 8, Figures 8 and 9), we find that the process
presents less variability in values of δ = 1.5, which suggests a group of hoppers receive
content that is far away plus 1.5 standard deviations of the average content according to
the target weight and the type of product to be packed. Finally, the results suggest that for
k = 7, hopper combinations the packaging process achieves its optimal value.

9. Conclusions
The multihead weighing process is characterized by high product packaging perfor-
mance and accuracy. The optimization of this process guarantees material savings and
high levels of productivity in terms of the number of packages produced. In this document,
a new packaging algorithm and its corresponding optimization model for double-layered
upright and diagonal multihead weighers have been presented, considering actual weights
in the weighing hoppers and a predefined number of hoppers to be combined. The algo-
rithm was validated using three product feeding strategies (S1 , S2 , and S3 ) and different
numbers k of hopper combinations. To assess the performance of the process, two products
(Ravioli and Fusilli) were tested at different coefficients of proportionality (γ), and then
an experiment design was approached to establish a comparison between different factor
levels. The performance metrics used were the average weight of the packages produced,
the standard deviation of the packages produced, and the coefficient of variation of the
packages produced. Results revealed that the best filling configuration to reduce the pro-
cess variability is the S1 strategy, particularly with the five subgroups of hoppers, assigning
the largest number of hoppers to the group. In addition, the number of hoppers to be
combined at k = 7 offered the least variability in the total weight of the packages produced
in both processes. The study also concluded that process behavior or operation is better
at minimum values of γ (γ = 0.123). Finally, the diagonal machine offers a greater NC
for the weighing hoppers when selecting the package weight, which is reflected in lower
process variability.

Author Contributions: Conceptualization, R.G.-J., J.C.G.-D., A.D.P.-R.; Methodology, R.G.-J., J.C.G.-

D. and A.D.P.-R.; Validation, R.G.-J., J.C.G.-D. and A.D.P.-R.; Formal analysis, R.G.-J., J.C.G.-D. and
A.D.P.-R.; Investigation, R.G.-J., J.C.G.-D. and A.D.P.-R.; Data curation, R.G.-J.; Writing—original
draft preparation, R.G.-J.; writing—review and editing, R.G.-J., J.C.G.-D. and A.D.P.-R.; Visualization,
R.G.-J., J.C.G.-D. and A.D.P.-R.; Supervision, J.C.G.-D. and A.D.P.-R. All authors have read and agreed
to the published version of the manuscript.
Funding: This research received no external funding.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: Not applicable.
Mathematics 2021, 9, 1039 20 of 20

Acknowledgments: We express our gratitude for the support from Universidad Simón Bolivar, and
Universitat Politècnica de València.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Ishida Co., Ltd. Multihead Weighers. Available online: http://www.ishida.com/ww/en/products/weighing/ccw/ (accessed on
12 April 2021).
2. Yamato-Scale Co., Ltd. Filling and Packaging (Products). Available online: http://www.yamato-scale.co.jp/en/products/index
(accessed on 12 April 2021).
3. Montgomery, D.C.; Borror, C.M. Systems for modern quality and business improvement. Qual. Technol. Quant. Manag. 2017,
14, 343–352. [CrossRef]
4. Taguchi, G. Introduction to quality engineering, Tokyo. Asian Product. Organ. 1990, 4, 10–15.
5. Roy, R.K. A Primer on the Taguchi Method; Society of Manufacturing Engineers: Southfield, MI, USA, 2010.
6. Montgomery, D.C. Introduction to Statistical Quality Control, 8th ed.; John Wiley & Sons: New York, NY, USA, 2019.
7. Keraita, J.N.; Kim, K.-H. A study on the optimum scheme for determination of operation time of line feeders in automatic
combination weighers. J. Mech. Sci. Technol. 2006, 20, 1567–1575. [CrossRef]
8. Keraita, J.N.; Kim, K.H. A Weighing Algorithm for Multihead Weighers. Int. J. Precis. Eng. Manuf. 2007, 8, 21–26.
9. Karuno, Y.; Nagamochi, H.; Wang, X. Bi-criteria food packing by dynamic programming. J. Oper. Res. Soc. Jpn. 2007, 50, 376–389.
10. Karuno, Y.; Nagamochi, H.; Wang, X. Optimization Problems and Algorithms in Double-layered Food Packing Systems. J. Adv.
Mech. Des. Syst. Manuf. 2010, 4, 605–615. [CrossRef]
11. Imahori, S.; Karuno, Y.; Nagamochi, H.; Wang, X. Kansei engineering, humans and computers: Efficient dynamic programming
algorithms for combinatorial food packing problems. Int. J. Biomed. 2011, 3, 228–245. [CrossRef]
12. Imahori, S.; Karuno, Y.; Nishizaki, R.; Yoshimoto, Y. Duplex and Quasi-Duplex Operations in Automated Food Packing Systems.
In Proceedings of the IEEE/SICE International Symposium on System Integration (SII), Fukuoka, Japan, 16–18 December
2012; pp. 810–815.
13. Karuno, Y.; Takahashi, K.; Yamada, A. Dynamic Programming Algorithms with Data Rounding for Combinatorial Food Packing
Problems. J. Adv. Mech. Des. Syst. Manuf. 2013, 7, 233–243. [CrossRef]
14. Karuno, Y.; Saito, R. Heuristic algorithms with rounded weights for a combinatorial food packing problem. J. Adv. Mech. Des.
Syst. Manuf. 2017, 11. [CrossRef]
15. Beretta, A.; Semeraro, Q.; Del Castillo, E. On the Multihead Weigher Machine Setup Problem. Packag. Technol. Sci. 2016, 29,
175–188. [CrossRef]
16. Del Castillo, E.; Beretta, A.; Semeraro, Q. Optimal setup of a multihead weighing machine. Eur. J. Oper. Res. 2017,
259, 384–393. [CrossRef]
17. García-Díaz, J.C.; Pulido-Rojano, A. Monitoring and control of the multihead weighing process through a modified control chart.
DYNA 2017, 84, 135–142. [CrossRef]
18. Pulido-Rojano, A.; García-Díaz, J.C. Optimisation algorithms for improvement of a multihead weighing process. Int. J. Prod. Qual.
Manag. 2020, 29, 109. [CrossRef]
19. García-Díaz, J.C.; Pulido-Rojano, A.; Giner-Bosch, V. Bi-objective optimisation of a multihead weighing process. Eur. J. Ind. Eng.
2017, 11, 403. [CrossRef]
20. García-Díaz, J.C.; Pulido-Rojano, A. Performance analysis and optimisation of new strategies for the setup of a multihead
weighing process. Eur. J. Ind. Eng. 2020, 14, 58. [CrossRef]
21. Wishon, C.; Villalobos, J.R. Robust efficiency measures for linear knapsack problem variants. Eur. J. Oper. Res. 2016,
254, 398–409. [CrossRef]
22. Gao, C.; Lu, G.; Yao, X.; Li, J. An iterative pseudo-gap enumeration approach for the Multidimensional Multiple-choice Knapsack
Problem. Eur. J. Oper. Res. 2017, 260, 1–11. [CrossRef]
23. Avci, M.; Topaloglu, S. A multi-start iterated local search algorithm for the generalized quadratic multiple knapsack problem.
Comput. Oper. Res. 2017, 83, 54–65. [CrossRef]
24. Montgomery, D.C. Design and Analysis of Experiments, 9th ed.; John Wiley & Sons: New York, NY, USA, 2017.
25. Santos-Fernández, E. Multivariate Statistical Quality Control Using R; Springer: Berlin/Heidelberg, Germany, 2012.