See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/339027271

Development and Application of Kanban and Milk-Run in Production Process

of a Metalworking Company

Conference Paper · December 2019

DOI: 10.1109/IEEM44572.2019.8978753

CITATION READS

1 565

4 authors, including:

Juan Carlos Quiroz-Flores José C. Alvarez

Universidad de Lima Universidad Peruana de Ciencias Aplicadas (UPC)
149 PUBLICATIONS 203 CITATIONS 56 PUBLICATIONS 166 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Juan Carlos Quiroz-Flores on 07 October 2020.

The user has requested enhancement of the downloaded file.

 Development and Application of Kanban and Milk-Run in Production
Process of a Metalworking Company
A. F. Caballero-Barrera1, J. P. Valdivia-Castillo1, J.C. Quiroz-Flores1, J. C. Alvarez-Merino1
1
Ingeniería Industrial, Universidad Peruana de Ciencias Aplicadas (UPC), Lima, Perú
([email protected], [email protected], [email protected], [email protected].)

Abstract – Demand variability and high competitiveness assembly line involves evaluation of multiple factors. One
in the manufacturing sector have influenced companies to of them is the importance of just-in-time, as any delay of
adopt different tools to optimize their inventory control. The material replenishment can result in stoppage of the
Kanban tool offers a reduction of minimum inventory levels assembly line, which leads to inactive time for the workers.
and increased production flexibility. This study covers the
In contrast, if the materials are replenished too early, this
design, application, and validation of a Kanban control
system and a milk-run distribution system in a kitchen can lead to an accumulation of materials in the areas close
manufacturer company in Peru. The main problem faced by to the station, which tend to be narrow [4].
the company is the high index of work in progress (WIP)
generated by the bottleneck of the productive flow. This is While multiple studies have been performed on Kanban,
caused mainly by the absence of a flow system that allows most of them were focused on the determination of the
control of the production. This problem amounts to a monthly number of Kanban cards and the impact on the
cost of approximately $53,272. In this study, a pilot is performance of different card loop structures. Those
implemented in the production line to evaluate the results studies do not consider the logistics involved in making the
obtained and, in parallel, a simulation system is designed to
evaluate the long-term results. From the results obtained, it
system work [1]. Based on that, this can be classified into
can be concluded that the application of the proposed tools (i) constant order-quantity systems and (ii) constant order-
will reduce the WIP of each component produced, avoiding cycle systems [5] [6]
overproduction and optimizing costs.
In response, this study explores how the application of a
Keywords - production, metalworking, Kanban, lean, Kanban system, combined with milk-run in a single case
WIP, visual control, milk run, Peru study of a metalworking company can improve the
problem of the high level of work in progress (WIP).
I. INTRODUCTION
The remainder of this paper is organized as follows.
The metalworking sector covers a wide variety of Section 2 presents the case study. Section 3 presents the
industries that produce a broad range of products [1]. This methodology and the applied proposed model. Results are
sector continued a growing trend in recent years, becoming presented in Section 4. Section 5 provides the conclusions.
the second sector with the highest sales revenue in the
manufacturing industry in Peru [2]. According to global II. CASE STUDY
indicators, the annual growth rate of the household
appliance industry increased by 4.7% [3]. However, the The case study was conducted in a company that
participation of Latin America in this industry only produces kitchens. This research will be focused on the
represents 2.8%. Therefore, it is of crucial importance for production of four kitchen burners, as this was the principal
companies to maintain their competitiveness, to operate product according to the analysis.
efficiently, optimizing their use of human resources,
machinery, time, etc. The main problem encountered was the high level of
inefficiencies in the cutting and pressing areas, which is
The planning of the dispatch of material to an reflected in the WIP (Fig. 1). The latter was also

Fig. 1. Losses found in the current productive process

978-1-7281-3804-6/19/$31.00 ©2019 IEEE1250 1250

 Proceedings of the 2019 IEEE IEEM

identified as the bottleneck of the process in Fig. 2. The III. PROPOSED MODEL
main losses identified were due to the low availability, high
level of WIP, and high index of defective components. The proposed model consists of controlling the volume
of inventories in the process through application of the
Kanban tool, and the dispatch logistics through application
of milk-run. The components involved in the process are
shown in Figure 4.

Fig. 2. Takt time Vs cycle time (sec)

According to the cost analysis, the cost of having a high

level of WIP represents 62% of the total cost of the
problem.
Fig. 4. List of components

First, a calculation is performed using two techniques,

namely the XYZ analysis to identify the more consumed
components and subsequently the ABC analysis to find
those components that have lower cost. Based on these two
analyses, it is determined which components have a
production using in the Kanban system. The pilot will be
implemented on the most frequent and less costly
components.

The analysis of the calculation is determined with the

following parameters:
Fig. 3. Percentage of cost of each loss factor
1) Daily consumption of two previous months
The Kanban tool in a decentralized system in inventory 2) Daily consumption planning for two months
control regulates the production of necessary products in 3) Average consumption
the essential quantities and in the time required [7]. 4) Standard deviation of consumptions
Moreover, the Kanban system allows the process flow to 6) Maximum daily consumption
be optimal, since it reduces bottlenecks [8]. The classical 7) Security stock for each material
methodology of the system is carried out by means of cards 8) Frequent materials with XYZ analysis
for the control of the production flow [9], i.e., it controls 9) Valorized materials with ABC analysis
the inventory at each stage of the process [10]. When all indicated parameters are found, the
components are classified in Kanban or production order,
In contrast, according to Silva et al, [11] many considering that the components with frequency X and
investigations on the Kanban system only focus on the some of Y must be produced by Kanban.
constant order quantity and not on the constant order cycle.
For this reason, this study focuses on the milk-run as a In the next step, the number of cards of each color is
complement to the Kanban tool. calculated, and equations (1), (2), and (3) are used to this
end.
The milk-run system is described as a cyclic material ‫ݐ݋ܮ‬ (1)
͓ ‫ ݏ݀ݎܽܿ ݊݁݁ݎܩ‬ൌ
handling system based on routes that allow frequent ܰ‫ݎ݁݊݅ܽݐ݊݋ܿݎ݁݌ݏ݁ܿ݁݅݌݂݋ ݎܾ݁݉ݑ‬
deliveries in containers, as needed from central storage, to
௅௘௔ௗ ௧௜௠௘‫כ‬஼௢௡௦௨௠௣௧௜௢௡ (2)
multiple points [12]. In the study conducted by Bocewicz ͓ ܻ݈݈݁‫ ݏ݀ݎܽܿ ݓ݋‬ൌ
ே௨௠௕௘௥ ௢௙௣௜௘௖௘௦௣௘௥௖௢௡௧௔௜௡௘௥
et al. [13], problems are indicated in the selection of routes
that supply the order, and implementing milk-run ܵ݁ܿ‫݊݋݅ݐ݌݉ݑݏ݊݋ܥ כ ݇ܿ݋ݐݏ ݕݐ݅ݎݑ‬ (3)
minimizes downtime and production time. ͓ ܴ݁݀ ܿܽ‫ ݏ݀ݎ‬ൌ
ܰ‫ݎ݁݊݅ܽݐ݊݋ܿݎ݁݌ݏ݁ܿ݁݅݌ ݂݋ ݎܾ݁݉ݑ‬

978-1-7281-3804-6/19/$31.00 ©2019 IEEE1251 1251

 Proceedings of the 2019 IEEE IEEM

TABLE I IV. VALIDATION

COLOR OF KANBAN FOR EACH COMPONENT

Validation of the proposed improvement has been

Number of cards pieces of total of carried out in two ways, through the implementation of a
Components for Kanban
container cards pilot and through simulation of the system in the Arena
LATERAL EXT GRIS L60 5 5 3 188 13 Software.
PISO DEFLECTOR INF GA 1 1 1 1656 3
MESA 3M+1G C/ VENA IN 3 2 1 300 6 A. Validation through a pilot
FDO PTA HNO EK2 L60 3- 3 2 1 300 6
FRENTE HNO EK2 L60 3-3 3 3 2 300 8 The implementation of the pilot has been developed
PISO HNO EK2 L60 3-3 (OP 1 2 1 1120 4 for the three components with the greatest amount of
ESPALDAR DE HORNO EK 2 2 1 720 5 consumption (“lateral,” “piso,” and “mesa”). This required
PTA ESTUFA EMB COLDE 3 3 2 150 8 a blackboard where the cards and containers could be seen,
PANEL DE CTRL IX C/TRA 3 3 2 150 8 and where the components will be placed (Fig. 6).

In the case of component dispatch, the milk-run system

is implemented, for which the path of each component
must be identified. We use the number of containers
previously found in the Kanban to determine the time
during which the container will be used (Equation 4).

‫݁݉݅ݐ݊݋݅ݐ݌݉ݑݏ݊݋ܿݎ݁݊݅ܽݐ݊݋ܥ‬ሺ݄‫ݏݎݑ݋‬ሻ  (4)
‫݊݋݅ݐ݌݉ݑݏ݊݋ܿݕ݈݅ܽ݀݉ݑ݉݅ݔܽܯ‬
ൌ ‫ͳʹ כ‬Ǥ͹ͷ݄‫ݏݎݑ݋‬
ܰ‫ݎ݁݊݅ܽݐ݊݋ܿݎ݁݌ݏ݁ܿ݁݅݌݂݋ݎܾ݁݉ݑ‬

Moreover, it is essential to know the travel time from

point “x” to point “y”, as it must be considered in the
calculation to establish the distribution schedules of each Fig. 6. Kanban Pilot
component.
The results obtained during the pilot month are shown
The company before the improvement makes the in Table I, a significant decrease in the amount of WIP
following route to supply the different areas (Fig. 5), and presented in the pressing area can be observed.
this flow does not have an order or a frequency, such that
in one shift it can increase within an order of magnitude in TABLE II
each of the areas. PILOT RESULTS

It is important to highlight that for the proposed tools

to be implemented and developed in an optimal way, prior
training needs to be provided to the operators regarding the
new system. In addition, a procedure has been carried out
such that the operational and administrative staff can
comply with this change in the types of programming.

In the case of the dispatch of components, the new

route was established through the milk-run tool (Fig. 7).
The results obtained are shown in Table III.

Fig. 5. Route of dispatch of components

978-1-7281-3804-6/19/$31.00 ©2019 IEEE1252 1252

 Proceedings of the 2019 IEEE IEEM

The simulation represents the production and

consumption of the components in the process. First, the
production has as variable the maximum amount that must
be manufactured according to the calculation, that is, the
number of containers per stock of each of them. Second,
consumption has as its main variable the security stock.
The flow begins when the process provides the maximum
number of components, then, when the consumption makes
the components reach the minimum batch, the system is
Fig. 7. Proposed route with milk-run system ordered to produce up to the maximum that is indicated
according to each material. In summary, when the stock
TABLE III reaches the minimum indicated, the software immediately
RESULTS OF MILK-RUN PILOT starts producing until it reaches the maximum stock.

Unlike the pilot, the simulation was performed taking

into account all components. The results obtained are
shown in Table V.

TABLE V
RESULTS OF SIMULATION BY COMPONENT

Inventory
B. Validation through simulation Component
Level (unit)

The Arena Simulation software was used to simulate Lateral externo 1677
the flow of production and consumption in the pressing Piso deflector 3800
process (Fig 8). The simulation was made considering three Mesa 1069
shifts of 8 hours a day each, during the course of a year.
Fondo puerta de
Herein, the evaluation of multiple variables was necessary, 1080
horno
namely the exogenous, controllable, and non-controllable Frente de horno 1316
variables.
TABLE IV Piso de horno 2413
SIMULATION VARIABLES Espaldar de horno 1820
Exogenous variable Puerta estufa 787
Inventory in process level panel de control 787
Controllable variables
Size of container
Production quantity
Reorder point
Non-controllable variables
Demand

Fig. 8. Simulation in Arena Software

978-1-7281-3804-6/19/$31.00 ©2019 IEEE1253 1253

 Proceedings of the 2019 IEEE IEEM

V. ANALYSIS OF RESULTS International Journal of Engineering Research and

Technology, 11(11), pp. 1713-1734.
The results obtained from the pilot (Table II) show that [5] Monden, Y. Toyota Production System: Practical Approach
the WIP% of the three analyzed components are 9% on to Production Management. New York, NY: Norcross:
Industrial Engineering and Management Press, 1983.
average, which represents a decrease of 14% compared to [6] Protzman, C., G. Mayzell, and J. Kerpchar. Leveraging Lean
the WIP level before the improvement. Regarding the milk- in Healthcare: Transforming Your Enterprise into a High
run, it is shown that the total time devoted to the dispatch Quality Patient Care Delivery System. 1st ed. New York:
of components per shift is reduced by 38% (Table III). Productivity Press, 2010.
[7] Andrade, Y., Cardenas, L., Viacava, G., Raymundo, C.,
Finally, an analysis of the results obtained by the Arena Dominguez, F., “Lean manufacturing model for the
Software simulation is carried out. The average WIP reduction of production times and reduction of the returns of
obtained is 13%, a value that is still considered acceptable. defective items in textile industry”, Advances in Intelligent
Systems and Computing, 954, pp. 387-398.
[8] Becerra, A., Villanueva, A., Núñez, V., Raymundo, C.,
Further, the time used for dispatch of components per Dominguez, F., “Lean manufacturing model in a make to
shift was reduced by 38%. order environment in the printing sector in Peru”, Advances
in Intelligent Systems and Computing, 971, pp. 100-110.
In general, the results of the proposal indicate that this [9] H. A. Lima Rocha et al., “Bed management team with
method manages to reduce the level of inventory in the Kanban web-based application,” Int. J. Qual. Heal. Care,
process. vol. 30, no. 9, pp. 708–714, 2018.
[10] A. Ang and R. Piplani, “A performance comparison between
TABLE VI the Extended Kanban Control System (EKCS) and the
FUTURE STATE OF METRICS Traditional Kanban Control System (TKCS),” in
IEEM2010-IEEE International Conference on Industrial
Expected Pilot Simulation Engineering and Engineering Management, 2010.
As is [11] C. Silva, L. M. Ferreira, M. Thürer, and M. Stevenson,
% WIP 23% 10% 9% 13% “Improving the logistics of a constant order-cycle kanban
system,” Prod. Plan. Control, vol. 27, no. 7–8, pp. 650–659,
Jun. 2016.
VI. CONCLUSIONS [12] Y. A. Bozer and D. D. Ciemnoczolowski, “Performance
evaluation of small-batch container delivery systems used in
The present study has demonstrated that the lean manufacturing-Part 1: System stability and distribution
application of Kanban and milk-run tools have a positive of container starts,” Int. J. Prod. Res., vol. 51, no. 2, pp. 555–
effect on inventory in process indexes (WIP). This means 567, 2013.
that less storage space is used, thus reducing storage and [13] G. Bocewicz, P. Nielsen, and Z. Banaszak, “Declarative
modeling of a milk-run vehicle routing problem for split and
maintenance costs.
merge supply streams scheduling,” in Advances in
Intelligent Systems and Computing, 2019, vol. 853, pp. 157–
Investment necessary for the implementation of the 172.
project is $5,438, which is much lower than the monthly
cost of the problem before the improvement ($53,272).
Thus, it can be concluded that the project is economically
viable.

For the improvement to be applied properly, it is

important to work on the organizational culture. Therefore,
it is advisable to work in a transversal manner to encourage
the commitment of both operators and the administrative
staff.

REFERENCES

[1] Di Natale, Picón, Quezada and Toro, “Planeamiento

Estratégico del Sector Metalmecánico en el Perú,” M.S.
thesis, CENTRUM Pontificia Universidad Católica del Perú,
Lima, Perú, 2017.
[2] Promexico, "Diagnostico Sectorial. Electrodomésticos,”
Unidad de Inteligencia de Negocios, Mexico, 2016.
[3] ProMexico, “Industria Electrodomésticos,” Secretaría de
Economía, México, 2014.
[4] Guillen, K., Umasi, K., Quispe, G., Raymundo, C., “LEAN
model for optimizing plastic bag production in small and
medium sized companies in the plastics sector”, (2018),

978-1-7281-3804-6/19/$31.00 ©2019 IEEE1254 1254

View publication stats