Solving the Bottleneck Problem in a Warehouse Using SimulationsActa Logistica
Solving the Bottleneck Problem in a Warehouse Using SimulationsActa Logistica
doi:10.22306/al.v8i2.209 Received: 05 Dec. 2020; Revised: 26 Jan. 2021; Accepted: 20 Feb. 2021
~ 107 ~
Copyright © Acta Logistica, www.actalogistica.eu
Acta logistica
logistica - International Scientific Journal about Logistics
Volume: 8 2021 Issue: 2 Pages: 107-116 ISSN 1339-5629
correct selection of employees, their motivation and based on ExtendSim was researched by [23] and machinery
qualifications, also correct assignment of employees to allocation in the container terminal by [24]. In the
workplaces and a fair distribution of work among the nonindustrial area, as an example can be mention [25],
employees is important. Personnel scheduling has a which used CPLEX and ExtendSim to solve the stochastic
significant influence not only on the warehouse optimisation model for Emergency Department.
productivity but also on the development of employee The aim of the paper is optimisation of the workforce
competencies [12]. Ernst et al. [13] considered an allocation in the warehouse. Simulation and optimisation
employee scheduling problem arising in service industries are performed by software tool ExtendSim9. The
with flexible employee availability and flexible demand. In simulations will verify alternatives for assigning workers
the literature review [14], new trends and approaches were to the picking and packing workplace to eliminate the
presented for the personnel scheduling problems. existing bottleneck and uneven workload.
Similarly, authors of [15] reviewed rostering problems in
specific application areas and the models and algorithms 2 Materials and Methods
that had been reported in the literature for their solution. 2.1 Case study description
It is important to state that many software tools are The distribution warehouse of the internet storage with
developed for staff scheduling, often based on nutritional supplements is operated as an e-shop without
mathematical models using efficient optimal or heuristic stone shops. Customers order goods through the company's
algorithms. Masoud [16] developed a simulation-based website. The store delivers the goods within 24 hours, if
optimisation framework for labour management to the order was placed within the specified time limit. If later,
optimise labour allocation. Andersson [17] presented a then the shipment is delivered within 2 days.
simulation-optimisation system for personnel scheduling Approximately 80% of orders go to export. The average
in which the multi-objective evolutionary algorithm number of orders received is 2 600 to 2 800 per day. Table
NSGA-II had been implemented. Among the software 1 shows the time and number of received orders from
tools, ExtendSim is often used for simulation and 15.01.2020, when 2 673 orders were received. Picking
optimisation [18]. In the literature, there are presented starts at 6:00 and lasts until 18:00. From the overview in
many researches and real-word studies about ExtendSim Table1, it is evident that most orders come for picking at
application. For example, using ExtendSim to optimise the 6:00, but this does not mean that customers sent their order
delivery process is presented in [19], to streamline at this time. All orders that arrive after the 24-hour delivery
production logistics in [20], to optimise wastes flow in [21] time limit are carried over to the next day and packed in the
and assess the meat processing and cutting production in morning at the beginning of the shift.
[22]. Also, optimising the command and control process
Table 1 Time and number of orders received for individual transport companies
Transport company /Number of orders
Time Cargus GLS Hr Zásielkovňa PPL GLS HU Post DHL
6:00 236 42 83 73 116 95 76
7:00 7 6 10 8 11 7 6
8:00 37 10 18 14 18 14 13
9:00 58 17 26 21 34 19 19
10:00 67 22 31 37 38 30 27
11:00 73 27 43 43 44 39 34
12:00 87 32 46 46 57 37 41
13:00 17 6 54 55 63 51 56
14:00 69 54 51
15:00 26 39 62
16:00 12 21 20
17:00 6 16
Picking up goods in the warehouse is done manually The company uses services of seven transport companies
using human force. Orders are packed according to for delivery. Transport companies and departure times for
priority. The highest priority is given to orders that belong individual transport companies are listed in Table 2.
to the transport company that leaves the warehouse first.
~ 108 ~
Copyright © Acta Logistica, www.actalogistica.eu
Acta logistica
logistica - International Scientific Journal about Logistics
Volume: 8 2021 Issue: 2 Pages: 107-116 ISSN 1339-5629
~ 109 ~
Copyright © Acta Logistica, www.actalogistica.eu
Acta logistica
logistica - International Scientific Journal about Logistics
Volume: 8 2021 Issue: 2 Pages: 107-116 ISSN 1339-5629
3 Results and discussion 3.1 Proposal No. 1 with the available number of
The simulation model is applied to experimentally employees
verify the operation of the warehouse and optimise the The first proposal simulates a system with an available
schedule of activities for the workforce. Two proposals for number of employees, representing 28 employees per one
the organisation of work are verified experimentally. shift. The model is first necessary to enter the measured
Proposal number 1 considers the current number of values for the individual "Activity" blocks. In the first
employees and experimentally redistributes the employees "Activity" block, which represents Picking, we set the
at both workplaces in order to find the optimal distribution shortest time to 2.5 minutes, the longest time to 5.5 minutes
of the workforce. Proposal number 2 envisages an and the most likely time to 4 minutes. The second
increased number of workers to reduce the system's high "Activity" block represents the packaging lines, where the
workload and tension. Also, in proposal number 2, the time settings are as follows: shortest: 2.5 minutes, longest:
schedules of the increased number of employees are 4 minutes and the most likely time to 3.3 minutes. In the
experimentally verified in order to achieve an optimal "Transport" block, a value of 0.6 minutes for the move time
workload utilisation of approximately 80%. This would is entered, representing the time required to move the crate
reduce the risk of bottlenecks and failure of processes. from Picking to the packing line. Finally, it is necessary to
enter the number of employees in the individual "Activity"
blocks. The utilisation of "Activity" blocks is shown in
Figure 3.
~ 110 ~
Copyright © Acta Logistica, www.actalogistica.eu
Acta logistica
logistica - International Scientific Journal about Logistics
Volume: 8 2021 Issue: 2 Pages: 107-116 ISSN 1339-5629
Figure 4 and Figure 5 display the course of the colour curves. The vertical axis expresses the number of
simulation in proposal 1. In the figures, order packing processed orders. The graphs show that all orders received
times for individual transport companies are expressed by were packaged but just before the time limit.
Figure 4 Result of proposal No. 1 for transport companies (Cargus, GLS HR, Zásielkovňa , PPL)
Cargus, GLS HR, Zásielkovňa, PPL
~ 111 ~
Copyright © Acta Logistica, www.actalogistica.eu
Acta logistica
logistica - International Scientific Journal about Logistics
Volume: 8 2021 Issue: 2 Pages: 107-116 ISSN 1339-5629
Figure 5 Result of proposal No. 1 for transport companies (GLS HU, Post, DHL)
GLS HU, Post, DHL
Experimenting with the deployment of employees, we 3.2 Proposal No. 2 with an increased number of
came to the conclusion that the optimal solution is 15 employees
people on Picking and 13 people on Packing. The The second proposal simulates a system with an
advantage of this proposal is that the company does not increased number of employees. The setting of times in
have to spend any funds to hire more employees. The individual "Activity" blocks is the same as in proposal 1.
disadvantage is the high workload, which represents 98% In this proposal, we have increased the number of
at the Picking and 93% at the Packing (Figure 3). Another employees at Picking by 3 and at the Packing by 2. A total
disadvantage is the small time reserve, which means that of 33 workers are employed in this proposal. The result is
the system operates in a highly tense mode. From the a reduction in Picking utilisation from 98% to 82% and
graphs in Figure 4 and Figure 5, we can see that all orders Packing from 93% to 80%. The utilisation of "Activity"
were packed but just before the time limit. An unexpected blocks in proposal 2 is shown in Figure 6.
increase in orders would mean that the warehouse would
not be able to ship all orders, which would lead to non-
delivery of the order and cause customer dissatisfaction.
~ 112 ~
Copyright © Acta Logistica, www.actalogistica.eu
Acta logistica
logistica - International Scientific Journal about Logistics
Volume: 8 2021 Issue: 2 Pages: 107-116 ISSN 1339-5629
Figure 7 and Figure 8 show the course of the simulation and the vertical axis shows the number of orders. The
in proposal 2, where the colour curves represent the graphs show that the time reserves are much larger than in
packing times of orders for individual transport companies proposal 1.
Figure 7 Result of proposal No. 2 for transport companies (Cargus, GLS HR, Picking, PPL)
Cargus, GLS HR, Zásielkovňa, PPL
~ 113 ~
Copyright © Acta Logistica, www.actalogistica.eu
Acta logistica
logistica - International Scientific Journal about Logistics
Volume: 8 2021 Issue: 2 Pages: 107-116 ISSN 1339-5629
The advantage of proposal 2 lies in the reduction of the increase of orders by about 300. The disadvantage of this
workload of employees and in the creation of larger time proposal is that the company would have to hire another 5
reserves. The graphs and Table 3 show that the time employees, which would result in an increase in labour
reserves have increased significantly compared to the first costs.
proposal. Such system can withstand an unexpected
Figure 8 Result of proposal No.2 for transport companies (GLS HU, Post, DHL)
GLS HU, Post, DHL
Table 3 Packaging times of the last package for information on the optimal distribution of staff, which
individual transport companies in proposal No.2 eliminated the bottleneck. The total number of employees
Transport 28 was in the optimal solution divided into 15 people for
Packing time of last order
company picking and 13 people for packaging. In this mode, the
Cargus 13:06 system managed to meet the requirements, all orders
managed to be packed on time. However, the workload of
GLS HR 13:18 picking workers was 98% and packaging 93%. The
Zásielkovňa 13:40 advantage of this proposal is that the company does not
PPL 13:50 have to spend any funds to hire more employees. The
disadvantage is the high work commitment and small time
GLS HU 16:10 reserves, which would, in the event of an unexpected
Post 17:12 increase in orders, cause the warehouse not to be able to
DHL 17:20 ship all orders.
Proposal 2 considered an increased number of
employees. The aim of this proposal was to remove the
The analysis of activities in the warehouse showed that bottleneck and at the same time reduce the workload of
the company has a problem in the organisation of work in workers to about 80%. In case of unexpected delays and
the warehouse, specifically a bottleneck in the picking sudden increases in orders, this burden creates enough
section. The problem in the organisation of work was
reserve time. Repeated experiments on the simulation
caused by the uneven distribution of employees in the
model have provided a solution for the optimal distribution
picking and packaging departments. To overcome this
of workers in increased numbers. The results of the
problem, two proposals were verified by ExtendSim9
simulation showed that if we increase the number of
simulation program to find the optimal allocation of workers in picking by three and in the packaging by two,
workforce into workplaces.
the utilisation rate in picking will decrease to 82% and in
Proposal 1 considered the currently available number packaging to 80%. In such a work schedule setting, the
of workers, while the simulations examined the
system would be able to withstand an unexpected increase
possibilities of dividing the workers into the picking and in orders of approximately 300.
packaging department. The simulation results provided
~ 114 ~
Copyright © Acta Logistica, www.actalogistica.eu
Acta logistica
logistica - International Scientific Journal about Logistics
Volume: 8 2021 Issue: 2 Pages: 107-116 ISSN 1339-5629
By comparing both proposals only based on direct motivation, skills, and qualifications of the workforce.
costs, proposal 1 is a more appropriate solution to the Also, the work environment and workplace equipment
bottleneck problem. It does not require any additional affect the performance provided. These factors have not
costs, and by a simple organisational change, the been included in the presented simulation model.
bottleneck can be eliminated. In terms of the overall risk of
the assessed system, indirect costs, and concerning Acknowledgement
sustainability, the solution according to proposal 1 is This article was supported by the state grant agency KEGA
unacceptable, or acceptable only in the short term. The 012TUKE-4/2019, VEGA 345 1/0638/19, APVV-19-0418 and
analysed company is therefore forced to seek a solution to APVV SK-SRB-18-0053.
the bottleneck by increasing capacity. This is either by
increasing the number of own employees, by relocating References
workers from other workplaces, if possible, or by hiring [1] AMORIM-LOPES, M., GUIMARÃES, L., ALVES, J.,
labour force for exposed periods. Another possibility of ALMADA-LOBO, B.: Improving picking performance
eliminating the bottleneck is investing in the modernisation at a large retailer warehouse by combining probabilistic
of the used technologies. simulation, optimisation, and discrete-event
simulation, International Transactions in Operational
4 Conclusions Research, Vol. 28, No. 2, pp. 687-715, 2021.
Capacity constraints determine system performance. doi:10.1111/itor.12852
Simulation programs are often used in solving capacity [2] DERHAMI, S., SMITH, J. S., GUE, K. R.: A
problems and bottlenecks elimination. The article dealt simulation-based optimisation approach to design
with the issue of bottleneck elimination in the distribution optimal layouts for block stacking warehouses,
warehouse using computer simulation. The initial analysis International Journal of Production Economics,
of the processes identified the bottleneck at one of the Vol. 223, 2020. doi:10.1016/j.ijpe.2019.107525
warehouse workplaces (packing), which led to the risk of [3] SEBO, J., BUSA, J.: Comparison of Advanced
non-compliance with the requirements for sending orders Methods for Picking Path Optimisation: Case Study Of
within the time limit. The reason for the creation of the Dual-Zone Warehouse, International Journal of
bottleneck was preliminarily determined by the unbalanced Simulation Modelling, Vol. 19, No. 3, pp. 410-421,
distribution of warehouse workers to workplaces. Based on 2020. doi: 10.2507/IJSIMM19-3-521.
the determination of time and capacity parameters of the [4] TEPLICKÁ, K., ČULKOVÁ, K.: Using of optimising
process, a simulation model was created in the ExtendSim9 methods in inventory management of the company,
program and the warehouse process was simulated. Acta logistica, Vol. 7, No. 1, pp. 9-16, 2020.
Experiments with a different redistribution of employees doi:10.22306/al.v7i1.150
were used to find the optimal solution so that the process is [5] MIRČETIĆ, D., RALEVIĆ, N., NIKOLIČIĆ, S.,
managed, and workload allocation is optimal. MASLARIĆ, M., STOJANOVIĆ, Đ.: Expert System
The simulations resulted in two proposals for staffing Models for Forecasting Forklifts Engagement in a
the Picking and Packing sites. In the first proposal, the Warehouse Loading Operation: A Case Study,
current number of workers was considered 28, the optimal PROMET - Traffic&Transportation, Vol. 28, No. 4, pp.
solution was to put 15 on picking and 13 on packing. 393-401, 2016.
However, based on simulations, this solution would mean [6] SINGH, M., ARDJMAND, E.: Carton Set
mastering the process (all orders are sent on time), but the Optimization in E-commerce Warehouses: A Case
workload of employees would be 98% (picking) and 93% Study, Journal of Business Logistics, Vol. 41, No. 3,
(packing) and even with a small increase in orders, the pp. 222-235, 2020. doi:10.1111/jbl.12255
process would be unmanageable. In the second solution, an [7] JIANG, Z. Z., WAN, M., PEI, Z., QIN, X.: Spatial and
increased number of employees was considered and in temporal optimisation for smart warehouses with fast
order to achieve not only the mastery of the process but turnover, Computers and Operations Research,
also the optimal load of workplaces (approximately 80%). Vol. 125, 2021. doi:10.1016/j.cor.2020.105091
This solution requires increasing the number of workers for [8] STOPKA, O., ĽUPTÁK, V.: Optimisation of
picking by 3 and for packing by 2. warehouse management in the specific assembly and
The advantage of the used method is solving the distribution company: A case study, Nase More,
problem on the virtual simulation model and verification of Vol. 65, No. 4 Special issue, pp. 266-269, 2018.
the impacts of the proposed solutions in advance and doi:10.17818/NM/2018/4SI.19
practically with no financial costs. But several limitations [9] PRŮŠA, P., JOVČIĆ, S., SAMSON, J.,
also need to be pointed out. The simulation model is still a KOZUBÍKOVÁ, Z., KOZUBÍK, A.: Using a non-
simplification of reality and does not take into account parametric technique to evaluate the efficiency of a
many, even significant factors. Work productivity and logistics company, Transport Problems, Vol. 15, No. 1,
workplace performance depend not only on the number of pp. 153-161, 2020. doi:10.21307/TP-2020-014
employees and the production volume but also on the
~ 115 ~
Copyright © Acta Logistica, www.actalogistica.eu
Acta logistica
logistica - International Scientific Journal about Logistics
Volume: 8 2021 Issue: 2 Pages: 107-116 ISSN 1339-5629
[10] SHETTY, N., SAH, B., CHUNG, S. H.: Route [19] MATINDI, R., HOBSON, P., MASOUD, M., KENT,
optimisation for warehouse order picking operations G., LIU, S. Q, "Developing a versatile simulation,
via vehicle routing and simulation, SN Applied scheduling and economic model framework for
Sciences, Vol. 2, No. 2, 2020. doi:10.1007/s42452- bioenergy production systems, International Journal
020-2076-x of Industrial Engineering Computations, Vol. 10, No.
[11] BAHRAMI, B., AGHEZZAF, E. H., LIMERE, V.: 1, pp. 17-36, 2019. doi:10.5267/j.ijiec.2018.5.003
Using Simulation to Analyse Picker Blocking in [20] STRAKA, M., KHOURI, S., ROSOVA, A.,
Manual Order Picking Systems, Procedia CAGANOVA, D., CULKOVA, K.: Utilisation of
Manufacturing, Vol. 11, pp. 1798-1808, 2017. computer simulation for waste separation design as a
doi:10.1016/j.promfg.2017.07.317 logistics system, International Journal of Simulation
[12] DENKENA, B., DITTRICH, M. A., WINTER, F., Modelling, Vol. 17, No. 4, pp. 583-596, 2018.
WAGENER, C.: Simulation-based planning and doi:10.2507/IJSIMM17(4)444
evaluation of personnel scheduling in knowledge- [21] MIKUŠOVÁ, N., TOMKOVÁ, E., DOVICA, M.,
intensive production systems, Production DEBELIC, B., PERIC-HADZIC, A., ZAJAC, J.: Use
Engineering, Vol. 10, No. 4-5, pp. 489-496, 2016. of Simulation for Waste Management and Reverse
doi:10.1007/s11740-016-0693-4 Material Flow, Advances in Science and Technology
[13] AĞRALI, S., TAŞKIN, Z. C., ÜNAL, A. T.: Research Journal, Vol. 12, No. 4, pp. 136-143, 2018.
Employee scheduling in service industries with doi:10.12913/22998624/94965
flexible employee availability and demand, Omega, [22] PLÀ-ARAGONÉS, L. M., PAGÈS-BERNAUS, A.,
Vol. 66, No. 4, pp. 159-169, 2017. NADAL-ROIG, E., MATEO-FORNÉS, J.,
doi:10.1016/j.omega.2016.03.001 TARRAFETA, P., MENDIOROZ, D., PÉREZ-
[14] ÖZDER, E. H., ÖZCAN, E., EREN, T.: A Systematic CÀNOVAS, L., LÓPEZ-NOGALES, S.: Economic
Literature Review for Personnel Scheduling Assessment of Pig Meat Processing and Cutting
Problems, International Journal of Information Production by Simulation, International Journal of
Technology and Decision Making, Vol. 19, No. 6, pp. Food Engineering, Vol. 16, No. 5-6, 2020.
1695-1735, 2020. doi:10.1142/S0219622020300050 doi:10.1515/ijfe-2018-0100
[15] ERNST, A. T., JIANG, H., KRISHNAMOORTHY, [23] GE, B., XIA, B., YANG, Z., ZHAO, Q., WEI, H.:
M., SIER, D.: Staff scheduling and rostering: A Modeling and analysis of ExtendSim model and data-
review of applications, methods and models, driven command and control processes, Xi Tong
European Journal of Operational Research, 2004, Gong Cheng Yu Dian Zi Ji Shu/Systems Engineering
Vol. 153, No. 1, pp. 3-27. doi:10.1016/S0377- and Electronics, Vol. 42, No. 5, pp. 1063-1072, 2020.
2217(03)00095-X doi:10.3969/j.issn.1001-506X.2020.05.13
[16] MASOUD, S., SON, Y. J., KUBOTA, C., [24] DUAN, Z., SONG, B.: Research on simulation
TRONSTAD, R.: Evaluation of simulation-based optimisation of container terminal's machinery
optimisation in grafting labor allocation, Applied configuration, Xitong Fangzhen Xuebao / Journal of
Engineering in Agriculture, Vol. 34, No. 3, pp. 479- System Simulation, Vol. 28, No. 6, pp. 1461-1468,
489, 2018. doi:10.13031/aea.12487 2016.
[17] ANDERSSON, M., SYBERFELDT, A., NG, A., [25] ALLIHAIBI, W., MASOUD, M., CHOLETTE, M.,
BENGTSSON, V.: 'Evolutionary simulation- BURKE, J., KARIM, A., LIU, S. Q.: 'Optimising the
optimisation of personnel scheduling' 12th service of emergency department in a hospital',
International Industrial Simulation Conference 2014, Proceedings - 22nd International Congress on
ISC 2014, 2014, pp. 61-65, 2014. Modelling and Simulation, MODSIM 2017, pp.
[18] FEDORKO, G., MOLNAR, V., HONUS, S., 1255-1261, 2017.
NERADILOVA, H., KAMPF, R.: The application of doi:10.36334/modsim.2017.i2.allihaibi
simulation model of a Milk Run to identify the
occurrence of failures, International Journal of
Simulation Modelling, Vol. 17, No. 3, pp. 444-457, Review process
2018. doi:10.2507/IJSIMM17(3)440 Single-blind peer review process.
~ 116 ~
Copyright © Acta Logistica, www.actalogistica.eu