Available online at www.sciencedirect.

com

ScienceDirect
Procedia Manufacturing 8 (2017) 59 – 66

14th Global Conference on Sustainable Manufacturing, GCSM 3-5 October 2016, Stellenbosch,
South Africa

Modelling, simulation and optimization of the materials flow of a

multi-product assembling plant
Wilson R. Nyembaa,*, Charles Mbohwab
a
Department of Mechanical Engineering Science, Faculty of Engineering and the Built Environment, University of Johannesburg, Auckland Park
2006, Johannesburg, South Africa
b
Professor of Sustainability Engineering, Department of Quality and Operations Management & Vice Dean for Research and Innovation, Faculty
of Engineering and the Built Environment, University of Johannesburg, Auckland Park 2006, Johannesburg, South Africa

Abstract

Various dynamic factors impact the movement of materials within a manufacturing environment, increasingly becoming complex
for multi-product assembling plants owing to the multiplicity and interconnectedness of these factors. Analyzing these factors can
be equally complex, requiring modelling and simulation tools. This paper looks at the modelling and simulation of the materials
flow of a multi-product furniture assembling plant to develop an efficient system that accomplishes timely product deliveries at
minimal cost. Generic simulation models based on 2 products were developed and constructed using Arena® Simulation Software.
Following the simulation experiments and implementation of the results, the average hourly throughput was significantly increased
and additional space to store materials prior to processing at workstations was created. The generic models were compatible with
the company's other products and hence useful for the company’s production planning and scheduling.

©©2017
2016TheTheAuthors. Published
Authors. by Elsevier
Published B.V.B.V.
by Elsevier This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the organizing committee of the 14th Global Conference on Sustainable Manufacturing.
Peer-review under responsibility of the organizing committee of the 14th Global Conference on Sustainable Manufacturing

Keywords: Arena®; materials flow; modelling; optimisation; simulation; workstation

* Corresponding author. Tel.: +263 772 345 441; fax: +263-4-303280.

E-mail address: [email protected] or [email protected]

2351-9789 © 2017 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the organizing committee of the 14th Global Conference on Sustainable Manufacturing
doi:10.1016/j.promfg.2017.02.007
60 Wilson R. Nyemba and Charles Mbohwa / Procedia Manufacturing 8 (2017) 59 – 66

1. Introduction

The key objective of any manufacturing company is to maximize on profit margins, hence the production function
of manufacturing companies holds the key to the success of these organizations [1]. The case study in this research is
a furniture manufacturing and timber processing company in which production flows are characterized by many
dynamic and complex factors which usually result from the unpredictable manner in which orders are placed let alone
the dynamic flow of materials in the factory, often leading to the failure by the company to meet customer orders on
time. This was evidenced by the variability of the average number of products produced per week over a period of 3
years from 2009 to 2012. Manufacturing companies have realized the need to invest in state-of-the-art and modern
methods of manufacturing coupled with computer integrated manufacture for accuracy, consistence and repeatability
[2]. Most indigenous owned manufacturing companies in Zimbabwe evidently lagged behind in the acquisition of
modern equipment and systems owing to limitations in financial capacity. This was exacerbated by the global recession
that affected many developed and emerging economies from around 2008, especially in Sub-Saharan Africa because
of weak global linkages [3]. Local companies in Zimbabwe at that time focused on production, leaving very little or
no funds for research or new techniques. In the absence of proper working systems and production schedules, a lot of
the work was being carried out haphazardly, resulting in delays in meeting customer orders. Following an ‘as-is-
analysis’, other problems identified included; old and obsolete equipment, thus creating bottlenecks in production,
backtracking and crisscrossing of process flows, resulting in long distances travelled by parts and contributing to
delays in production, hence costly products. Simulation models can be built to study the effectiveness of different
forms of materials handling equipment by considering their detailed parameters such as speed, process paths and
control logic. This paper looks at the modelling and simulation of the company’s multi-product furniture
manufacturing and assembling plant with the objective of appropriately scheduling production using the right product
mixes and process flows through optimizing materials flow to accomplish timely product deliveries and thus achieve
sustainable operations. Most simulation models are not generic but specific for each system that is being studied.
Although the two models in this paper were developed for two products using a limited simulation software, they were
generic in that only slight modifications were required in order to experiment on other products at the company.

2. Background to modelling and simulation

Models to be simulated can represent a real-world process more realistically because fewer restrictive assumptions
are required [5]. Consequently, simulation provides a more realistic replication of the dynamic nature of the flow of
materials within a factory rather than to rely completely on static analysis, which can be misleading in establishing a
good system [6]. Production schedules, variation in product mixes, availability of materials handling equipment, and
random breakdowns create varying loads on the system [4]. Static and dynamic analysis should both be utilized in
evaluating the efficiency of a plant layout in terms of flow of materials for complete, accurate and timely analysis.
More essentially, the simulation approach does not disrupt the on-going activities on the factory setup but it provides
a problem identification and solving tool that is flexible and less costly than physical prototyping and experimentation.
The approach also allows time compression, whereby simulation accomplishes in minutes what might require years
of actual experimentation. One way of accomplishing timely product deliveries is through designing an efficient
materials flow system by modelling and experimenting on product mixes depending on orders that would have been
received and thus assisting in production scheduling. Modelling and simulation normally starts with the proper
identification of the problem which entails specification of objectives and identification of the relevant controllable
and uncontrollable variables of the system to be studied [7, 15]. Due to the nature and complexity of simulation to
problem-solving, it should really be used as a last resort, after ascertaining that other approaches such as queuing
theory cannot be used to solve the particular problem, a vital aspect in the initial stages of modelling and simulation.
[8, 9]. The first step in constructing a simulation model is determining which properties of a real system should be
fixed (parameters) and which should be allowed to vary throughout the simulation run (variables). Variables for
models are specified by either of the two categories of distributions used for simulation; empirical frequency
distributions and standard mathematical distributions. Such distributions have to be determined by direct observation
or detailed analysis of records but other situations can reasonably be assumed to closely approximate a standard
mathematical distribution such as normal or Poisson [10, 16]. The length of the simulation runs depend on the purpose
 Wilson R. Nyemba and Charles Mbohwa / Procedia Manufacturing 8 (2017) 59 – 66 61

of the simulation. The most common approach is to run the simulation for a set period, such as a month, to see if the
conditions at the end of the period appear reasonable. Errors may arise in the program from mistakes in coding or in
logic [11, 17]. The input data to a simulation model is obtained through work measurement and this includes process
flows, distances between workstations, mode of materials handling and time to perform an operation [8]. Simulations
are primarily concerned with experimentally predicting the behavior of a real system for the purposes of designing or
modifying its behavior to achieve a certain purpose or solve a particular problem [12]. In conducting a simulation
experiment, values for the controllable inputs are selected and values for probabilistic inputs are randomly generated.
Based on these values, the model is then used to compute the values of the output [13]. However, simulation is not an
optimization technique but a tool that can be used to predict how a system will operate given certain conditions for
the controllable inputs and randomly generated values for the probabilistic inputs [14]. Quantitative analyses often
make use of simulation to determine values for the controllable inputs that are likely to tend towards desirable outputs,
in which case, simulation can be an effective tool for designing systems to provide good performance. Depending on
the simulation results, attempting additional runs on the experiments may be ideal and this can be achieved by
changing factors such as parameters, variables, decision rules, starting conditions and run length [2].

3. Case study and methodology

The case study company specializes in mid-volume or batch production and assembly of domestic and commercial
furniture as well as industrial timber products. In selecting the products for use in the modelling and simulation
research and eventual construction of the generic simulation models, consideration was given for those products that
went through most of the workstations as well as being representative of the manufacturing processes at the company.
Pallets were chosen to represent the generic simulation model for industrial timber products while baby tenders were
chosen to represent the generic simulation model for domestic furniture. Fig. 1(b) illustrates the ten stage material
flow for baby tenders in which the flow starts with the arrival of raw timber in the timber yard, hereinafter Workstation
1 (WS1) followed by the selection of suitable timber for the various parts of the baby tenders before being ferried
through 54m to the surfacer (WS2) for planing and throughout the illustrated stages until WS10. Pallets go through
an almost similar but shorter process as shown in Fig. 1(a).

3.1. Process flows

(a) (b)
Timber Raw Timber Selection Timber Raw Timber Selection
Yard (WS1) Yard(WS1)
79.5m Cross-Cutter Cross-Cutting 54m Surfacer (WS2/5) 3m Planing
(WS2) 14m Lamination Laminating
Parts 11m Rip Saw (WS3) Ripping (WS3)
Multi-Cutter 9m Moulding 13m Thicknesser Sizing
(WS4) (WS4)
Parts

Pallet 26m Rip Saw

Assembly Assembling (WS6)
(WS5) Cross- 10m
Cutter(WS7) Cross-
Cutting
Spindle 4.5m
Moulder Moulding
assembly

(WS8)
Sub-

Drum 28m
Sander Sanding
(WS9)
Nursery 43.5m
Assembly Assembly and Final Fitting
(WS10)

Fig. 1. (a) Five stage process flow for pallets; (b). Ten stage process flow for baby tenders
62 Wilson R. Nyemba and Charles Mbohwa / Procedia Manufacturing 8 (2017) 59 – 66

3.2. Mathematical model

The generic simulation models for both the pallets and baby tenders were developed based on Figs. 1(a) and (b) in
conjunction with the materials flowchart using Arena® Simulation Software. The 4 time parameters used were:

at material movement time from previous workstation to active workstation.

bt waiting time before being processed at the active workstation.
ct product processing time at the active workstation.
dt waiting time after processing before moving to the next workstation

The size of a component or product is an important consideration in assembly line analysis and design because the
number of products that can be handled at each workstation affects worker performance. Hence the number of products
n, entering the assembly line at a given time should not exceed a certain value and thus n became one of the controllable
inputs to the simulation model while the other were the number of workstations, w. The probabilistic inputs to the
materials flow simulation model were the 4 durations, at, bt, ct and dt derived from data collection and analysis at the
same plant [19]. The output consisted of various operating characteristics such as average queueing time, hourly output
and the total time spent in the assembly line as shown in Fig. 2 along with equations (1) and (2) for the computation
of total time spent at a workstation, tw and in the system ts. After developing the general simulation model outlining
the inputs and outputs of the model, this was extrapolated further into developing the materials flow chart which
defined the sequence of logical and mathematical operations required to simulate the materials flow for pallets and
baby tenders, taking note of the appropriate processing, idle and movement times.

Probabilistic Inputs
at bt ct dt

Controllable Inputs
Model for computing
Products, n time taken by n
Workstations, w products
n
tw ¦ (a
x 1
t  bt  ct  d t ) x (1)
Total time spent in
the system, ts
s
ts ¦t
w 1
w (2)

Total time Operating

In System Characteristics

Fig. 2. General simulation model for an s-stage flow line.

Owing to the complexity of real production processes, a number of assumptions were made prior to developing the
models to provide a representative picture of the operations and outcomes. These included, two continuous 8 hour
shifts in a day with minimal or no breakdowns, assuming machines would run uninterrupted and any maintenance or
breakdown attended to during the off-shift periods. However in reality this may not be possible as machines can
breakdown at any time. The models also assumed that orders received from retailers follow a uniform pattern based
on previous sales history [21]. Due to scarcity of broad data, more experiments were required for validation.

3.3. Machine operating characteristics and parameters of the models

Workstations, process and assembly lines were identified as standard features of the generic simulation models.
The flow system consisted of a number of workstations, five for the pallets and ten for the baby tenders. In addition,
each workstation had two important resources; a machine to perform a task and a storage space to store components
 Wilson R. Nyemba and Charles Mbohwa / Procedia Manufacturing 8 (2017) 59 – 66 63

prior to processing. Each machine was assigned 4 possible states namely; starved (idle), busy (processing), blocked
(storage space full) or failed (broken down). In the constructed models, the failure state was a function of a machine’s
average uptime and downtime percentages. The failure state was dependent on the company’s breakdown maintenance
strategy. The storage space capacity for each workstation was also assigned three possible states namely; empty,
partially full or full capacity. Upon arrival at a workstation the components were delayed by a duration of time
specified by a standard mathematical distribution, equivalent to the waiting time before processing. After processing,
the material was delayed before being transported to the next workstation, depending on the status of the next
workstation and this was also specified by a standard mathematical distribution. Four parameters identified and used
in the two simulation models were; machine capacity (parts that can be handled at a time), storage space capacity for
each workstation, total number of workstations in each model and machine uptime and downtime percentages. These
are shown in Tables 1(a) and (b).

Table 1. Parameters for the simulation models (a) industrial – pallets; (b) domestic – baby tenders
(a) Parameter (b) Parameter
Work Machine Uptime Downtime Workstation Work Machine Uptime Downtime Workstation
Station Capacity % % Capacity (Parts Station Capacity % % Capacity (Parts
/ Workstation) / Workstation)
WS1 1 100 0 24 WS1 1 100 0 16
WS2 1 75 25 24 WS2 1 75 25 12
WS3 1 75 25 24 WS3 1 100 0 12
WS4 1 80 20 24 WS4 1 70 30 8
WS5 1 100 00 8 WS5 1 75 25 12
WS6 1 80 20 10
WS7 1 65 35 10
WS8 1 70 30 18
WS9 1 60 40 14
WS10 1 100 0 8

3.4. Variables, verification and validation of the simulation models

The variables for the simulation models were obtained from previous work on data collection and statistical data
analysis in conjunction with optimization of the plant layout and materials handling system of the plant [19, 20]. These
variables were custom built probability distributions of the various time periods that characterized materials flow for
the pallets and baby tenders and as defined in section 3.2. The probability distributions with their shape parameters
for the Gamma (γ), Beta (β), Normal (η) and Uniform (μ) distributions and the mean for the Exponential (ε)
distribution (in braces) for these durations are shown in Tables 2(a) and (b). A warm up period of about 3600 seconds
was used as the starting condition for both models to discard any bias in the initial phases of the simulation run. The
approach adopted in running the program was to perform the simulations continuously until an equilibrium condition
was achieved when there were minimal variations in output results. The number of observations required to attain
equilibrium were set in the simulation programs as the terminating conditions of the simulation runs, the run length
being determined by the duration required to achieve an equilibrium.

Table 2. Variables and their probability distributions for the simulation models; (a) industrial – pallets; (b) domestic – baby tenders
(a) (b)
Work Variables and Probability Distributions Work Variables and Probability Distributions
Station at bt ct dt Station at bt ct dt
WS1 - - γ{3,1} β{0.0,8} WS1 - - γ{3,1} γ{2,1}
WS2 ε{23.03} Β{3,1.5} γ{3,1} β{2,0.8} WS2 γ{3,1} β{1,2} γ{2,1} γ{2,1}
WS3 ε{14.95} γ{3,1} γ{3,1} β{1.5,3} WS3 β{3,1.5} β{3,1.5} γ{3,1} β{2,2}
WS4 ε{114.15} γ{3,1} β{3,1.5} γ{3.1} WS4 γ{3,1} γ{2,1} β{3,1.5} β{3,1.5}
WS5 β{2,0.8} γ{3,1} ε{271.8} γ{3,1} WS5 β{2,0.8} ε{213.88} β{1.5,3} ε{262.329}
WS6 γ{3,1} β{1.5,3} γ{2,1} γ{3,1}
WS7 β{3,1.5} β{3,1.5} γ{3,1} β{3,1.5}
WS8 η{52.8,1.5} γ{3,1} γ{3,1} μ{16.5,313.2}
WS9 γ{3,1} β{1,1} γ{3,1} β{3,1.5}
WS10 γ{3,1} γ{3,1} β{3,1.5} β{3,1.5}
64 Wilson R. Nyemba and Charles Mbohwa / Procedia Manufacturing 8 (2017) 59 – 66

Verification and validation were carried out to check whether the models were a valid translation of the flowchart
and representative of the factory’s materials flow. Challenges encountered included mistakes in coding by trying to
run the program and let the computer review the errors for correction. This was resolved by splitting the program into
small sections, followed by checking the validity of each section. Quantitative results from the models were compared
with the raw data collected from the factory to verify the validity of each model. The experience gathered during the
‘as-is-analysis’ at the plant and the knowledge of operations for the company assisted a great deal and played a major
part in the validation process. In a related research, probability distributions were obtained for the company’s four
main products of which results derived for bunk beds and baby tenders were used as input to the generic simulation
models for this paper [19]. The sample sizes were initially an average of 25 readings recorded through each
workstation in an hour and this figure increased slightly to 37 per hour after implementing recommendations on
product mixes from this research. The validity of the models gave management an appreciable level of confidence
because of the fairly reasonable representation of the real system based on the data and probability distributions. Fig.
3 (a) and (b) show snapshots of the generic simulation models for pallets and baby tenders developed in Arena®.
(a) (b)

Fig. 3. Snapshots of the simulation models in Arena® (a) industrial model - pallets; (b) domestic model - baby tenders).

4. Results

Qualitative results were obtained from the direct observation of materials flow in the generic models through which
flow bottlenecks were revealed as well as those workstations which were mainly; starved, busy, with large queues or
with high frequencies of breakdowns. The generic simulation model for pallets showed that there were queues of
materials at all workstations, with the largest occurring at WS2, WS3 and WS5 while shorter queues were observed
at WS1 and WS4. Machines at WS2 and WS3 failed more frequently while those at WS1 and WS5 hardly experienced
any failure. The generic simulation model for baby tenders showed that there were large queues at WS1, WS2, WS3
and WS4 while WS5 had a relatively shorter queue and no queues at all at the other 5 workstations. Machines at WS2,
WS6, WS7, WS8 and WS9 were observed to have the highest frequency of breakdowns. These observations and
qualitative results were useful to advise the company on issues such as, the use of preventive maintenance on those
machines that frequently broke down, consider investing in additional machines to augment those that had large
queues or disposal of old machines that frequently broke down and thus delaying production.
Quantitative results were an output from the simulation models which showed the average time spent by a product
in the system, average hourly throughput and average times spent by machines being starved, busy, blocked or broken
down (failed). The results gave the average times spent during production with the storage space being empty, partially
full or full. Tables 3 (a) and (b) show the qualitative results extracted from the Arena® simulation outputs for the
pallets and baby tenders respectively. The qualitative results tally with the quantitative observations indicating some
validity of the models. The quantitative and qualitative results from both models formed a good basis for which
optimization of the materials flow system was accomplished.
 Wilson R. Nyemba and Charles Mbohwa / Procedia Manufacturing 8 (2017) 59 – 66 65

Table 3. Average queue times for (a) industrial – pallets; (b) domestic – baby tenders
(a) (b)
Work Average Minimum Maximum Work Average Queue Minimum Maximum Value
Station Queue Time Value Value Station Time (Sec) Value (Sec) (Sec)
(Sec) (Sec) (Sec) WS1 1136.9 259.89 2767.2
WS1 4245.6 3754.2 19529 WS2 2243 893.51 3557.2
WS2 2624.6 404.3 6396.76 WS3 1686.9 509.06 3557.2
WS3 2635.6 484.59 6422 WS4 1675.7 499.59 3566.9
WS4 2005.5 1765.3 8904.6 WS5 65.294 0 1286.9
WS5 2648.8 485 6453.1 WS6 4.909 0 179.36
WS7 15.7 0 213.77
WS8 12.306 0 209.39
WS9 20.059 0 358.05
WS10 0.0518 0 6.6705

5. Discussion and recommendations

The approach taken in optimizing the materials flow system for the company was based on the results obtained
from the two generic simulation models. Emphasis was placed on reducing the time spent by a product in the system
to improve the system’s hourly throughput. Averages times for; product at a workstation, machines in all states,
queueing and a workstation’s space being empty, partially full or full were employed to establish an optimal materials
flow system. Results from the industrial simulation model showed that most of the production time and queueing was
at WS5 which was busy most of the time. Thus attention was needed at WS5 to ensure that the hourly throughput was
increased. This was probably because it was an assembly area where most of the work was carried out manually and
recommendations were made to automate the processes. Results from the domestic simulation model showed that
most of the production time was spent in the first 5 workstations where large queues were observed. This was attributed
to the slow pace at which WS5 operated, resulting in materials from WS4 being blocked, filtering down to WS3 and
so on. Additional machines were added at WS5 but it only improved the flow slightly as the throughput increased by
only one product, hence it would not be cost effective. Further analysis and experimentation showed that in general,
workstations 1-5 were much slower than anticipated. An identical block of machines was introduced to process
materials in parallel with machines at workstations 1-5 resulting in sufficient supplies of materials to workstations 6-
10 to prevent starving. Another option pursued was the creation of additional storage at workstations that had large
queues to prevent blockage of flow from downstream workstations. This minimized the frequencies at which upstream
workstations blocked materials from downstream ones, resulting in the average hourly throughput from 25 to 37
products while the average time spent by products in the queue decreased to 8541 from 10033 seconds. The average
hourly throughput was also increased significantly by doubling the number of workers in the assembly area.
Analysis by simulation and modelling showed that the company could turn around its fortunes and increase its
product throughput by reorganizing the process flows and thus fulfilling customer orders on time and ultimately result
in sustainable operations and company growth. Implementation of the results from this research also meant that the
company no longer resorted to the costly option of employing workers overtime. The simulation models constructed
in this research were made flexible for extension to other products. This allowed the company to analyse and validate
new products prior to launching them. The simulation approach used in this research has advantages of not only
improving the materials flow but also to experiment on a real system without disrupting the ongoing activities in the
factory. It therefore provided a fairly reasonable measure for predicting performance and planning for production
through experimenting on what-if scenarios which were used to decide on product mixes, what products and when to
produce them. The constructed simulation models also improved the company’s throughput for the company to realize
other benefits such as; reduced lead times, better utilization of space, equipment and machines and reduction in work
in process. However, the models were limited in size and in some instances had to be broken down because of the
limited version used. While this approach provided the company with a low cost option for planning and prediction
compared to physical experimentation, there were a number of assumptions made as highlighted in section 3.2, which
not only presented challenges for validation and verification but also some level of doubt in management confidence
to embrace the technology. Future work was recommended with a commercial version of Arena® that can handle
higher volumes of data and larger models.
66 Wilson R. Nyemba and Charles Mbohwa / Procedia Manufacturing 8 (2017) 59 – 66

6. Conclusions

Two fairly robust and generic simulation models were developed for the case study company’s main operations in
domestic furniture and industrial timber products. The constructed models were useful in optimizing the materials
flow for the company. Implementing some of the recommendations such as introducing parallel machine blocks,
additional workers in the manual assembly areas as well as creating additional storage space for materials prior to
processing helped in the flow of materials among workstations and the average hourly throughput was significantly
increased. The results from the research were welcomed by management and implemented immediately especially in
view of the fact that the options did not require capital investment in machinery but just competent systems analysis
in the use of the developed generic simulation models. The company managed to meet the demand for the company’s
main products of the pallets and baby tenders, delivering orders on time. The benefits realized from this research could
be invested in acquiring the commercial version of Arena® or other simulation software, additional machines for the
slow workstations and automating the manual assembly operations.

References

[1] E.M. Goldratt, J. Cox, The Goal – A process of ongoing improvement, third revised edition, North River Press, Massachusetts, 2004.
[2] R. Bloomfield, E. Mazhari, J. Hawkins, Y.J. Son, Interoperability of manufacturing applications using the Core Manufacturing Simulation Data
(CMSD) standard information model, J. Computers & Industrial Engineering, 62(2012), 1065–1079.
[3] S. Bakrania, B. Lucas, The impact of the financial crisis on conflict and state fragility in Sub-Saharan Africa, GSDRC Applied Knowledge Series
- Governance and Social Development Resource Centre Issues Paper, Available: http://www.gsdrc.org/go/emerging-issues#crisis, Accessed:
24 March 2016
[4] J. Wang, Q. Chang, G. Xiao, N. Wang, S. Li, Data driven production modeling and simulation of complex automobile general assembly plant,
J. Computers in Industry, 62(2011), 765–775.
[5] J.S. Baldwin, P.M. Allen, K. Ridgway, An evolutionary complex systems decision support tool for the management of operations, International
Journal of Operations & Production Management, 30(7), 2010, 700 – 720.
[6] N.V. Patel, T.E. Tariq, M. Khan, Theory of deferred action, Journal of Enterprise Information Management, 23(4), 2010, 521 – 537.
[7] W.R. Nyemba, T. Dzimba, Problem formulation and structuring for multi-criteria decision analysis of a wood processing company, Journal of
Science, Engineering and Technology, 1(1), 2013, 64-71.
[8] W.R. Nyemba, The role of modelling and simulation in decision-making for manufacturing enterprises, Proceedings of the Fifth International
Conference on Manufacturing Processes, Systems & Operations Management in Less Industrialized Regions, Bulawayo, 2002, 11-21.
[9] University of Warwick, Simulation of production systems, Warwick Manufacturing Group IGDS/MSc Module, 2001.
[10] R.B. Chase, N.J. Aquilano, F.R. Jacobs, Production and operations management- Manufacturing and services, eighth edition, McGraw-Hill,
New York, 1998.
[11] D.R. Anderson, D.J. Sweeney, T.A. Williams, J.D. Camm, J.J. Cochran, Quantitative methods for business, twelfth edition. South-Western
College Publishing, Cincinnati, 2013.
[12] F.S. Budnick, D. McLeavey, R. Mojena, Principles of operations research for management. AITBS Publishers, India, 1994.
[13] M. Rolón, E. Martínez, Agent-based modeling and simulation of an autonomic manufacturing execution system, J. Computers in Industry
63(2012), 53–78.
[14] A. Negahban, J.S. Smith, Simulation for manufacturing system design and operation: Literature review and analysis, Journal of Manufacturing
Systems, 33(2014), 241–261.
[15] V.S.K.V. Harish, A. Kumar, A review on modeling and simulation of building energy systems, J. Renewable and Sustainable Energy Reviews,
56(2016), 1272–1292.
[16] C. Antwerpen, N.J. Curtis, A data collection and presentation methodology for decision support: A case study of hand-held mine detection
devices, European Journal of Operational Research, 251(2016), 237–251.
[17] C.W. Yeh, D.C. Li, Y.R. Zhang, Estimation of a data-collection maturity model to detect manufacturing change, J. Expert Systems with
Applications, 39(2012), 7093–7101.
[18] J.P. Belaud, S. Negny, F. Dupros, D. Micheá, B. Vautrin, Collaborative simulation and scientific big data analysis: Illustration for sustainability
in natural hazards management and chemical process engineering, J. Computers in Industry, 65(2014), 521–535.
[19] W.R. Nyemba, C. Mbohwa, Data collection and statistical data analysis in preparation for simulation of a furniture manufacturing company,
Proceedings of the World Congress on Engineering, International Conference for Manufacturing Engineering and Engineering Management,
London, Vol. II, 2016, 727-732.
[20] W.R. Nyemba, C. Mbohwa, Optimization of a plant layout and materials handling system for a furniture manufacturing company, Proceedings
of the World Congress on Engineering, International Conference for Manufacturing Engineering and Engineering Management, London, Vol.
II, 2016, 832-837.
[21] W.R. Nyemba, C. Mbohwa, A new product development framework for a timber processing company, Proceedings of the World Congress on
Engineering, International Conference for Manufacturing Engineering and Engineering Management, London, Vol. II, 2016, 674-679.