The Factory Physics for the Scheduling: Application to
Footwear Industry
John Reyes
1
, Darwin Aldas
2
, Kevin Alvarez
1
, Mario García
1
and Mery Ruíz
2
1
Faculty of Engineering Systems, Electronics and Industrial, Universidad Técnica de Ambato,
Chasquis Avenue and Río Payamino Street, Huachi Campus, Ambato, Tungurahua, Ecuador
2
Faculty of Accounting and Auditing, Universidad Técnica de Ambato, Chasquis Avenue and Río Payamino Street,
Huachi Campus, Ambato, Tungurahua, Ecuador
Keywords: Productivity, Job Scheduling, Theory of Constraints, Work in Process, Cycle Time, Little Law, Promodel,
Process Simulation.
Abstract: In this paper, an analysis is made of variables such as inventory in process, cycle time and rate of production
of salable units proposed in the theory of constraints and applied in a case study in the manufacture of safety
footwear. The methodology used to evaluate these parameters is an experimental design with application of
Plant Physics laws in a simulation scenario with PROMODEL. The results reflect that the optimum quantities
of production for a cycle of work are achieved by adding 8.1% to the rate units that can be sold per unit of
time and reducing cycle time by 6.5%. It was determined that to produce a transfer batch the inventory in
process for each order hour is equivalent to 64% of the total of the units programmed for production based on
the maximum capacity of the manufacturing system.
1 INTRODUCTION
Effective production scheduling is essential for
successful operations that allocate resources over
time for specific tasks (Krajewski, et al., 2016), so an
unbalanced workload causes uncertainty in
organizations (Zhang & Wang, 2016).
In this context, optimal planning efficiency is
achieved by coordinating the production schedule
with Just in Time (JIT) distribution planning to
customers (Johar, et al., 2016), meeting deadlines,
optimizing resources, reducing time that does not add
value and increasing the production rate of the
system. By implementing operational planning,
scheduling, and production control strategies
organizations can make money today and in the future
(Goldratt & Cox, 2014).
Successful companies offer products and services
with optimal times of production and inventory
turnover, so three production management
approaches are important to achieve this goal:
Material requirements planning (MRPI, and MRPII),
Just in Time production and Theory of Constraints
(TOC).
TOC as a systematic management approach
focuses on eliminating the bottlenecks that prevent a
company's progress towards its goal of maximizing
profits and effectively utilizing the resources of the
production system (Mahapatra & Amit Sahu, 2006).
The productivity of a production system must
increase while its inventory and operating costs are
reduced, limited by the performance in the constraint
using TOC in the weakest link in the supply chain
(Günay, et al., 2014).
Factory physics provides a systematic description
expressed as laws of the underlying behavior of a
system. It helps to decide what performance measures
to collect and what alternatives to evaluate, as well as
in the interpretation of simulation results (Standridge,
2004).
To obtain a synchronized manufacturing process,
techniques are applied that analyze the constraints of
the production system, such as cycle time (CT),
which is the time a product spends since it enters the
system until it leaves, work in process (WIP) defined
by the inventory in process in the system and the
throughput (TH) referred to the rate units that can be
sold per unit of time. These techniques are based on
the approach proposed by Li, et al. (2005),
mentioning that the CT is an important indicator to
measure the performance of the factory. Sometimes it
is possible to lower the WIP level; however, excess
248
Reyes, J., Aldas, D., Alvarez, K., García, M. and Ruíz, M.
The Factory Physics for the Scheduling: Application to Footwear Industry.
DOI: 10.5220/0006403402480254
In Proceedings of the 7th International Conference on Simulation and Modeling Methodologies, Technologies and Applications (SIMULTECH 2017), pages 248-254
ISBN: 978-989-758-265-3
Copyright © 2017 by SCITEPRESS Science and Technology Publications, Lda. All rights reserved
WIP reduction can lead to decreased throughput.
Applying the Principles of Factory Physics, which
integrates Little's Law and the Operation Curve to
quantify the interdependence between operating
factors such as the Overall Equipment Efficiency
(OEE), CT, and WIP can be reduced the WIP level by
up to 10%.
Managing production scheduling has always been
a major challenge not only in the footwear industry
but in any field of production or service (Li, 2010).
Reyes, et al. (2016), when applying the TOC model,
proposes a management methodology for the
inventory of materials in the footwear industry, which
generates 18.7% of annual cost savings. However,
even today, footwear production processes continue
being totally artisan and the industries still follow the
conventional make to order production (Cocuzza, et
al., 2012).
Production planning and scheduling must evolve
and adopt new tools that help gain a deeper control of
what must be produced to efficiently meet customer
needs (Zangiacomi, et al., 2007). One of the tools
currently used to visualize the results that propose
strategies and methodologies of process
improvement, is the use of simulation software.
Simulation is a necessary tool to represent this
reality and make decisions that further reduce
manufacturing costs by validating the design or
redesign of any complex system such as footwear. In
this particular case the simulation is necessary for the
initial estimation of the throughput since a single
worker must perform multiple operations in several
workstations and the controls on the quantity of
inventory in process must be validated as well as the
costs of production (Grimard, et al., 2005).
The footwear industry presents WIP accumulation
caused by incorrect distribution of work areas,
inadequate production schedules and orders of
products not released on time (Marcelo, et al., 2016).
The accumulation of WIP generated by improper use
of job measurement techniques and erroneous sales
forecasts implies the execution of production orders
with long delivery times and orders that are not met.
The congestion of the productive system with the
presence of constraints finally reduces throughput
and business productivity.
This research proposes a study of the parameters
for shoe production through the detection of
manufacturing system constraints and their
evaluation with the laws of factory physics.
Using simulation with PROMODEL in a case
study, optimal results of TH, WIP and CT are
obtained with standard values of inventory in process
that an optimum shoe manufacturing system must
maintain to achieve the shortest cycle time and the
highest rate of production.
2 METHODOLOGY
The footwear industry in Ecuador has strengthened in
the last 6 years and generates in the country some
100,000 direct and indirect jobs, currently has about
5,000 Ecuadorian producers in the whole country,
being Tungurahua the province where concentrates
50% of the national production (Reyes Vasquez, et
al., 2016). From this sector of the country is taken as
case study an industry that produces safety footwear
because it is the product of greater demand.
For the study the corresponding manufacturing
processes are considered: punching, polishing,
trimming, fixing of eyelets, forming, assembly and
finished. The timing of the process is taken through
the chronometer technique with assessment of the
rhythm established in the leveling method proposed
in the Westinghouse system (Cevikcan & Kilic, 2016)
and calculation of supplements through a bi-objective
approach of assessment with fatigue standards
(Glock, et al., 2016). For the case study, the largest
overall time is taken for each of the processes to
encode in a Simulation model.
The timing and sequence of operations are entered
into the Promodel® process simulation software
version 7.0, which models the flow rate in the
manufacturing system, cycle times and the work in
process on a production line. Through simulated
experimentation it is proposes to validate the
hypothesis, which seeks to show that by keeping an
inventory in a constant process and reducing the cycle
time it is possible to increase the rate of production in
a situation in which the manufactured products go to
inventory. Reducing the size of process batches has a
much deeper effect than increasing the number of
transfer batches because the product mix is much
larger and generates reductions of WIP and waiting
times in production.
The parameters used in the simulation are
Throughput (TH) for machines, workstations,
production line and industrial installation per unit of
time, work in process (WIP) and cycle time (CT)
from the path between the beginning and end of
processes. One of the main restrictions of modeling is
to determine the bottleneck (Rb) which is the rate
(parts per unit time or jobs per unit time) of the
workstation that has the highest long-term utilization
(Hopp & Spearman, 2011).
In the Promodel simulation software, it is enter the
units that arrive to the system with a frequency
The Factory Physics for the Scheduling: Application to Footwear Industry
249
distribution of discrete random variable with
probability behavior of Poisson to obtain an
approximation to the reality of the manufacturing
system. This distribution is obtained with the
Promodel Stat Fit library comparing frequency
distributions. The experimentation is carried out
through simulation scenarios where the behavior of
the study variables is obtained.
For the evaluation of the relationship between the
productions parameters of plant physics, the Little
Law is used which establishes that there is a close
relationship between throughput (TH), cycle time
(CT) and work in process (WIP) defined by equation
(1). This equation shows that the longer the cycle time
will lower throughput with a constant level of
inventory. The research also seeks to obtain the
critical WIP of the line (Wo) which is the inventory
with the line produces the maximum throughput.
From this level, inventory only produces congestion
and storage costs. See equation (2) (Hopp &
Spearman, 2011)
TH = (WIP / CT) (1)
Wo = (rb) .To (2)
An adequate production program involves that
industries must adjust to the lower performance of the
rate of production. This is achieved by equations (3),
(4) and (5), where w is the WIP level.
TH best = w/To si w < Wo (3)
TH best = rb si w <Wo (4)
w= Wo (5)
The best performance law states that for a given
inventory level (w) and the highest throughput, are
obtained when the inventory level is equal to or
greater than the critical level (Wo). The result is equal
to the bottleneck rate (rb). The law of best
performance can also be formulated with equations
(6) and (7).
CT best= To si w <Wo o w= Wo (6)
CT best = w/rb, si w>Wo (7)
In order for companies to avoid developing their
manufacturing processes with the worst performance,
Plant Physics also evaluates this factor through
equations (8) and (9) for any level of inventory w.
TH worst= 1/To (8)
CT worst= w.To (9)
There is also the evaluation of the parameters
under a practical performance where an additional
time of CT by contingencies is considered. For this,
we use equations (10), (11), (12) and (13).
TH practical= (w/(Wo+w-1))rb (10)
CT practical= To+(w-1)/rb (11)
CT practical= ((w-1)/N) t + t (12)
CT practical = (1 + (w-1)/N) t (13)
The variable w represents the works in the
manufacturing system, N is the number of stations in
the production line, t refers to the process time in each
station, To = Nt is the system process time and rb = 1
/ T corresponds to the bottleneck. When a job arrives
at a station, the number of jobs waiting at this station
is (w-1) / N.
To the three cases proposed in Plant Physics, the
current case study is added, the equations for the CT,
TH and WIP of the four possible cases are modeled
in the Promodel® simulator: the best, the practical,
the worst and the current one. These results allow to
know the current state of the production system with
respect to the studied parameters in order to compare
through the generation of scenarios the variation of
the size of the order or the time of the simulation that
allows to obtain the greatest benefit for the
organization.
3 RESULTS
Figure 1: Distribution of probability of arrivals.
To program the attributes defined in the methodology
the Distributed Simulation (DS) method is used
(Anagnostou & Taylor, 2017). Locations are created
corresponding to each of the jobs where
manufacturing activities are developed and also 5
temporary storage areas that allow to know how many
units remain in process after a work shift.
The entity considered for the study is called
Model S15 which is the product in process. The
arrivals occur in defined trajectories from warehouse
of raw material to the production system with an
established frequency detailed in Figure 1. The
SIMULTECH 2017 - 7th International Conference on Simulation and Modeling Methodologies, Technologies and Applications
250
number of occurrences or units that arrive to the
system is infinite since in the first instance it is
simulated as a function of the time that is of 8 hours.
In total, the basic model generates 47 variables to
measure the key performance indicators (KPIs) such
as bottleneck rates (rb), process times (tp), cycle
times (CT), system efficiency, work in process WIP)
and trougthput (TH) for the three cases analyzed with
Little's Law (Hopp & Spearman, 2011).
5 networks of movements are traced in which the
14 operators cross each of the locations with a node
of the network through interconnections or interfaces,
in the displacements the distance measured in the
plant between the different jobs is taken into account.
Attributes of type real numbers are used for the entity
defined as att_S15, a subroutine named id is
programmed in order to avoid a division between zero
in the process, because the cycle time is only
calculated after the first one Part, this programmed
code is shown below:
IF CT_ModelS15 <>0 THEN
TH_ModelS15=wip_ModelS15/CT_ModelS15
ELSE
TH_ModelS15=0
To program the processes, you enter data such as:
input entity, output entity, input and output location,
route, movement logic.
It creates a variable Real x that is internal where
an operation time will be stored. The id subroutine
checks if the cycle times of each entity are different
from zero.
Then the program places the time that the clock of
the simulation is marking the moment the entity
enters the production system in attribute att_S15, the
equation to obtain the production capacity and the
respective bottleneck is also determined. A logic is
also proposed in order to obtain an increase in the
zones of temporary storage and at the same time a
decrease while they go on to the next process of the
system. Finally the equations are codified for the
analysis of the constraints with Little's Law, as shown
below:
REAL x
id
att_S15=CLOCK()
tp_1=1.30x=(tp_1)
rb1=1/x
WAIT x
INC WIP_Zona1
tp_2=1.90x=(tp_2)
rb2=1/x
WAIT x
DEC WIP_Zona1
IF rb1<rb2 AND rb1<rb3 AND rb1<rb4
AND rb1<rb5 AND rb1<rb6 AND rb1<rb7
THEN
{
TH_General= rb1
}
ELSE
IF rb2<rb3 AND rb2<rb4 AND rb2<rb5
AND rb2<rb6 AND rb1<rb7 THEN
{
TH_General= rb2
}
ELSE
IF rb3<rb4 AND rb3<rb5 AND rb3<rb6
AND rb3<rb7 THEN
{
TH_General= rb3
}
ELSE
IF rb4<rb5 AND rb4<rb6 AND rb4<rb7
THEN
{
TH_General= rb4
}
ELSE
IF rb5<rb6 AND rb5<rb7 THEN
{
TH_General= rb5
}
ELSE
IF rb6<rb7 THEN
{
TH_General= rb6
}
ELSE
TH_General= rb7
WIP_critical_best=TH_General*CT_Mode
lS15
INC WIP_critical_best
CT_Worst=wip_ModelS15*CT_ModelS15
CT_Practical=(CT_ModelS15+((WIP_Mode
lS15-1)/TH_General))
CT_Best= (WIP_ModelS15/TH_General)
TH_Practical=((WIP_ModelS15/(WIP_cri
tical_best+WIP_ModelS15-
1))*TH_General)
Capacity_best=TH_General*60
Capacity_practica=TH_Practical*60
Capacity_actual=TH_ModelS15*60
Production_best=Capacity_best*(CT_Mo
delS15/60)
production_practica=Capacity_practic
a*(CT_ModelS15/60)
Capacity_worst=TH_Worst*60
production_worst=Capacity_worst*(CT_
ModelS15/60)
Efficiency=(TH_ModelS15/TH_General)*
100
The Factory Physics for the Scheduling: Application to Footwear Industry
251
For movements between jobs is coded according
to the following:
MOVE WITH Operario1 FOR 2.28 THEN
FREE
With the data entered in the Simulator it is obtain
the bottleneck rate (rb) of the system of 0.53 pairs per
minute, as seen in Figure 2.
Figure 2: Bottleneck rate.
Table 1 shows the production parameters in the first
column and in the first row are located the four cases
analyzed from the theory of Plant Physics.
Table 1: Results of the four possible cases.
Best
Case
Worst
Case
Practical
Case
Curren
t Case
Units
WIP
238
1
238
238
Pairs
CT
449
11424
0
915,4
480
Minute
TH
0,53
0,002
0,26
0,49
Pairs/min
The time (To) consider for the simulation is 480
minutes equivalent to a working day of 8 hours, in this
time 238 pairs of shoes are produced with a
production rate of 0.49 pairs per minute.
With equation (2) it is obtain the critical Wip (Wo)
whose value corresponds to 254 pairs per day.
As expressed in equation (4), the best TH is found if
the rate of production reaches the value of rb,
therefore the value of TH = 0.53 pairs per minute is
taken for the best possible case. Equation (7)
calculates the best CT, this time is less than the
planning of a working day. With these results, there
was an increase of the TH by 8.1% and a reduction of
the CT by 6.5%.
The worst case is determined that in a work cycle
a pair of shoes is produced, with a production rate of
0.002 pairs / min and a cycle time of 114,240 minutes.
There is also a practical case that is calculated
with the equations whose production rate is 0.26 pairs
per minute where the company requires 915.4
minutes to fulfill the production of an order of 238
pairs.
This case is considered when there are unforeseen
for which the company must reduce its rate by 50%
and increase the time by 100%.
In order to determine the inventory in process that
the company must maintain in each hour of work, the
simulation is performed for a production of 50 pairs
of shoes since the workers produce in each
workstation in function of those quantities before
passing the product to the following process.
The Little Law that relates the production
parameters analyzes the current case, the best case,
the worst possible case and a practical case of
production for different WIP values, in order to verify
the dynamics of the process with different
parameters.
To schedule the production it is necessary to know
how the production flow values change over time and
also to know the maximum value of WIP within the
system in a set time so that there is no excess
inventory in process. Finally know how many units
can be produced in a given time.
Controlling inventories in process WIP also helps
to fulfill the delivery of orders on time in addition to
achieving an increase in productivity. It is necessary
to know in which time the highest TH is obtained and
how many units must be produced, for this the results
of the TH vs the CT of Figure 4 are observed.
The "Best Possible" case generates the minimum
cycle time and the maximum throughput for each
WIP level.
The production dynamics of the case study are
plotted in Figure 3 with the relationship between the
TH and the WIP; And Figure 4 the TH and CT in the
one-hour time cycle for the batch of 50 pairs, in order
to know how many units should be kept each hour
occurring in the system to meet the established goal
and the sequence of income Of the units to the system.
Table 2: Comparison of the four cases.
Case
WIP
(pairs)
TH
(pairs per min)
Worst
32
0,002
Best
32
0,53
Practical
32
0,26
Current
29
0,49
Table 2 shows that when a WIP of 32 pairs is in the
system, it is possible to reach the maximum
production rate, therefore it is the amount that the
plant must maintain every hour as a minimum to
fulfill the delivery of orders to time using the
maximum capacity of the production line.
After this value of 32 pairs, the cycle time is
increased in the time that is invested in producing a
SIMULTECH 2017 - 7th International Conference on Simulation and Modeling Methodologies, Technologies and Applications
252
0,000
0,100
0,200
0,300
0,400
0,500
0,600
0 20 40 60 80
TH
WIP
TH VS. WI P
TH BEST TH WORST
TH PRACTICAL TH CURRENT
0
50
100
150
200
250
300
0 20 40 60 80
CT
WIP
CT VS. WI P
CT BEST CT WORST
CT PRACTICAL CT CURRENT
pair with the best production rate. The results for the
other three cases are also presented.
Figure 3 shows that in the present case a
production rate of 0.49 pairs per minute with a CT of
60 min is achieved keeping an inventory in process of
29 pairs. According to the study, the company should
reach the best possible case by maintaining a
production rate of 0.53 pairs per minute with a CT of
60 minutes keeping an inventory in process of 32
pairs as shown in the green lines in Figure 3. From
these values the curve stops growing and the
inventory increases but not the production rate.
In the case study it is observed that in order to
maintain an inventory in process of 32 pairs the CT
must be increased to 123 min but the production rate
decreases to 0.26 pairs per minute.
Figure 3: Relation TH vs. WIP.
Figure 4: Comparison CT vs. WIP.
The results of Figure 4 indicate the variation of the
CT as a function of the WIP, in the best case possible
the time is kept constant in 60 minutes until an
inventory of 32 pairs. From this inventory the time
curve begins to grow which implies that to have more
units in the system the cycle time increases and
therefore the TH is reduced.
For every 50 pairs produced an inventory in
process of 32 pairs should be maintained every hour
within the system which implies that the inventory in
process should not exceed 64% of the total
production.
Figure 5 shows the behavior of the TH in the time
where the upper line shows the best possible case, it
is the only one that remains constant since it is the
case to which the system must tend optimally. The
second line indicates the current case with its
respective variability, this is due to different factors
that influence how the work rate of the workers or the
supply of raw materials, always must be taken care
that this line is between the best case and the practical
one to consider that the system of production is stable.
The third line is that of the practical case and the
fourth is the worst case possible.
Figure 5: Behaviour TH vs. CT.
4 CONCLUSIONS
A simulation model was programmed using the
Promodel software, considering the capacity
constraints and the production parameters where the
results were obtained for the four possible cases
stated in Plant Physics: best case, worst case, case
study and current case.
For a working day of 8 hours (480 minutes),
currently 238 pairs of shoes are produced with a
production rate of 0.49 pairs per minute.
Through the simulation it was determined that the
production rate that maximizes the capacity of the
line is 0.53 pairs per minute increasing from the
current one by 8.1%, also the reduction of the cycle
time For the best case to 449 minutes corresponding
this value to an improvement of 6.5% with respect to
the current one.
The Factory Physics for the Scheduling: Application to Footwear Industry
253
The inventory in process that must be maintained
at the maximum is 32 pairs per hour, which indicates
that if a greater value is conserved within the day, the
company begins to generate unnecessary
accumulation of the same one.
It is concluded that in order to maintain a
production rate that satisfies the maximum capacity
of the plant, every hour an inventory in process
equivalent to 64% of the total number of units
scheduled for production must be maintained.
For a manufacturing system to be optimal, the TH
production rate should not change over time but
remain constant, this is a clear indicator that resources
are coming to the system on time and supplying it is
efficient, where prevents waste generation.
The results of the model employed identify
companies with an efficient management of their
processes, and others that require support to analyze
and respond with data and facts scientifically proven
to eliminate the root causes of their problems.
ACKNOWLEDGEMENTS
The authors thank the National Footwear Chamber of
Ecuador and the Technical University of Ambato for
the support provided during the execution of this
work within the framework of the research project
called “Operational optimization based on a lean
dynamic system of alert of failures in the production
processes for the footwear industry”.
REFERENCES
Anagnostou, A. & Taylor, S., 2017. A distributed
simulation methodological framework for OR/MS
applications. Simulation Modelling Practice and
Theory, Volume 70, p. 101119.
Cevikcan, E. & Kilic, H. S., 2016. Tempo rating approach
using fuzzy rule based system and westinghouse
method for the assessment of normal time..
International Journal of Industrial Engineering, 23(1),
pp. 49-67.
Cocuzza, S., Fornasiero, R. & Debei, S., 2012. Novel
automated production system for the footwear industry.
In: Heidelberg, ed. IFIP Advances in Information and
Communication Technology (IFIPAICT). Berlin:
Springer, pp. 542-549.
Glock, . C. H., Battini, D., Grosse, E. H. & Alessandro, P.,
2016. Human energy expenditure in order picking
storage assignment: A bi-objective method. Computers
& Industrial Engineerin, Volume 94, p. 147157.
Goldratt, E. & Cox, J., 2014. The goal: a process of ongoing
improvement. 4th ed. Great Barrington: North River
Press.
Grimard, C., Marvel, J. & Standridge, C. R., 2005.
Validation of the re-design of a manufacturing work
cell using simulation. In Proceedings of the 37th
conference on Winter simulation , pp. 1386-1391.
Günay, N. S., Vayvay, Ö. & Şimşit, Z., 2014. Theory of
Constraints: A Literature Review. Procedia-Social and
Behavioral Sciences, Volume 150, pp. 930-936.
Hopp, W. J. & Spearman, M. L., 2011. Factory Physics.
Third ed. Long Grove: Waveland Press.
Johar, F., Nordin , S. Z. & Potts, C., 2016. Coordination of
production scheduling and vehicle routing problem
with due dates. In advances in industrial and applied
mathematics, 1750(1), p. 030035.
Krajewski , L., Malhotra, M. & Ritzman, . L., 2016.
Operations Management: Processes and Supply
Chains. 11 ed. Edinburgh: Pearson education.
Li, B., 2010. Research on the production scheduling
management system based on SOA. In: Lecture Notes
in Computer Science. Berlin: Springer, pp. 286-294.
Li, N., Zhang, L., Zhang, M. & Zheng, L., 2005. Applied
factory physics study on semiconductor assembly and
test manufacturing. In ISSM 2005, IEEE International
Symposium on Semiconductor Manufacturing, pp. 307-
310.
Mahapatra, S. & Amit Sahu, 2006. Application of Theory
of Constraints on Scheduling of Drum-buffer-rope
System. IOSR Journal of Mechanical and Civil
Engineering (IOSR-JMCE), pp. 15-20.
Marcelo, M. T. et al., 2016. Process improvement and
utilization of machines in the production area of a shoe
manufacturing company. In: Industrial Engineering
and Engineering Management (IEEM). Bali: IEEE.
Reyes Vasquez, J. P., Aldas Salazar, D. S., Morales
Perrazo, L. A. & García Carrillo , M. G., 2016.
Evaluation of the capacity for assembly in the footwear.
Ingeniería Industrial, 37(1), pp. 14-23 .
Reyes, J., Alvarez, K. & Vasquez, R., 2016. Dynamic
buffer management for raw material supply in the
footwear industry. Journal of Industrial and Intelligent
Information, 4(1), pp. 1-8.
Standridge, C. R., 2004. How factory physics helps
simulation. pp. 1103-1108.
Zangiacomi, A., Zhijian, L., Sacco, M. & Boër, C., 2007.
Process planning and scheduling for mass customised
shoe manufacturing. International Journal of Computer
Integrated Manufacturing, 17(7), pp. 613-621.
Zhang, J. & Wang, X., 2016. Multi-agent-based
hierarchical collaborative scheduling in re-entrant
manufacturing systems. International Journal of
Production Research, pp. 1-17.
SIMULTECH 2017 - 7th International Conference on Simulation and Modeling Methodologies, Technologies and Applications
254