Available online at www.sciencedirect.

com

ScienceDirect
Procedia Manufacturing 10 (2017) 100 – 111

45th SME North American Manufacturing Research Conference, NAMRC 45, LA, USA

Operations status and bottleneck analysis and improvement of a

batch process manufacturing line using discrete event simulation
Sriram Velumani, He Tang*
School of Engineering Technology, Eastern Michigan University, Ypsilanti, MI 48197, USA

Abstract

There are many product and process variations in the production systems of batch processing. Planning and executing such
manufacturing operations can be a significant and challenging task. Discrete Event Simulation is an effective tool to analyze the
moderate complex product and process variations and predict the operations status and bottlenecks of an existing manufacturing
system, which is critical to the operation planning, execution, and improvement. This study simulates the first process stage of
tire manufacturing in batch process involving product variation and process variants. The study analyses the operations status,
bottlenecks, and the interdependence of the manufacturing activities between machines. In the simulation modeling and analysis,
the efficiency of machines, reliability, quality, and setup time are considered. The simulation identifies the operation bottlenecks
and WIP status, and proposes process changes for improved production efficiency. The simulation study helps schedule
operations to reveal the requirement/necessity of changes or addition of the buffer based on buffer status analysis. The model can
also verify the changes of machines for throughput improvement.
© 2017 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
© 2017 The Authors. Published by Elsevier B.V.
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-reviewunder
Peer-review under responsibility
responsibility of the
of the Scientific
organizing Committee
committee of theof45th
NAMRI/SME.
SME North American Manufacturing Research Conference

Keywords: batch processing; discrete event simulation; bottleneck; buffer analysis; throughput improvement

1. Introduction

Batch processing is widely used in a high-volume production environment. Batch processing is planned and
executed with considerations of various product specifications and manufacturing processes to meet customer

* Corresponding author. Tel.: +1-734-487-2040; fax: +1-734-487-8755.

E-mail address: htang2@emich.edu

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 45th SME North American Manufacturing Research Conference
doi:10.1016/j.promfg.2017.07.033
 Sriram Velumani and He Tang / Procedia Manufacturing 10 (2017) 100 – 111 101

requirements. Compared with the repetitive processes in a high-volume production, the batch processing has more
variables and complex impacts on manufacturing operations. Therefore, to achieve optimal operational productivity
is challenging for the planning and execution of batch processes.
Discrete Event Simulation (DES) is an effective tool to analyze complex manufacturing operations with product
and process variability. In addition, DES has advantages for performing analysis to any future scenarios for
improved process time and efficiency. For example, DES has been used for the efficacy of introducing new
processes or equipment in an existing factory and measuring the interoperability. As a tool, DES is used for
imparting the data available with most of influential variables in the model, the product mix, at different stages and
sequences in this study.

1.1. Literature Review

The line configuration for implementing lean manufacturing in high volume production encompass the use of
pull systems (i.e., JIT, low inventories, short setup times, and make-to-order), emphasis on immediate resolution of
quality problems, cultivation of multi-skilled workers, and use of worker teams [1]. The systems versatility for
futuristic or newly-established production setups needs to be studied based on the design viewpoint of the
production layout designer which would indicate the follow-up processes need to be taken in time for example
change in layout, stand by machine setup and routing, repair arrangements and troubleshooting setups and resuming
production. There are also studies in complex manufacturing environment, such as shipyard, that are complicated to
be resolved by mathematical analysis and simulation methods have proven better for analysis and understanding
their behaviors [2].
The Little’s Law predicts material stagnation due to work-in-process (WIP) or machine failures leading to long
process time. The theory of constraints focus on the rigid bottleneck that causes the main inventory of the system.
The buffer line must be set up for the production to have smooth flow and keep pulling in the orders. There is
analytical equation available for buffer level calculation of the system with coefficient of variation and serial line
machines [3]. For a new or trial-run system, the analysis on the pre-optimized production plans or existing
production plans using multi objective criteria has also been studied in the basis of cost and bottleneck [4]. Their
study show that an optimization-based mathematical model is used with the material flow simulation. These models
also suggest that the subroutine models can be built in order to set objective parameters in the existing visual model
templates for optimization. The visual model templates being modeled correctly allows a good platform to analyze
the variation happening in the environment and be used for simulation-based forecasting and scheduling [5].
The use of analytical methods for different line configurations have been discussed [2, 3, 6]. For efficiency,
several criteria taken into consideration for lean enterprise auditing as in WIP, buffer status, bottlenecks
identification and are considered for improvement, such as smallest buffer capacity with multiple units sufficient to
ensure production rate. Analyzing the serial production machines with re-entrant lines through the aggregated
models with forward and backward aggregation are developed and compared with simulations [2].
Computer simulation is considered an indispensable method of problem solving [7] in different application
contexts, outstandingly necessary in the cases of testing of a complex system and design of a new system [8, 9]. The
simulation modeling approach can consider all decision variables and systems parameters, such as capacity and
machine’s speed of operation that affect the performance measures of the system such as production rate serve as
input [10, 11]. In all cases the system and inputs must be learned upon time for any high-volume production runs
alike initial inventory and minimum runtime. Most of the DES software has the capability to add routines such as
bottleneck detection methods that increases its potential [12].
The papers studied basic serial and parallel process lines with different efficiency, which provides a good
platform for the study of DES for different applications like shipyard and automotive manufacturing. The studies
established general guidelines for modeling, such as distribution selection, implementation of effective time for
processing and warm up times [13]. The bottleneck identification methods and their algorithm gives an insight in
how to interpret the results from the software. With these ideas and studies with essential parameters, the DES can
be applied to different manufacturing environment, for example complex batch processing. All the research and
literature provide a good foundation on a special study on multiple product and process variables in a batch process
model.
102 Sriram Velumani and He Tang / Procedia Manufacturing 10 (2017) 100 – 111

1.2. Research Problem Description

Tire manufacturing, as a typical process in a batch mode, is selected for this analysis. The tire manufacturing line
is a moderate complex process flow, which consists of five main machine units along with three final machine units,
as shown in Figure 1. Most of products to be produced have job orders, which have specific recipes or process steps
to meet the customers’ requirements.

Fig. 1. Manufacturing process flow

The production line has a combination of the two groups of machines: main and final. The main mixer and
convex mill roller machines are considered as a machine unit, that is, Main 1, Main 2, Main 3, Main 4, and Main 5.
The final machine mixer and four roller mills have the same flow of process considered as a machine unit Final 1,
Final 2, and Final 3.
In the process, the product materials enter from the hopers to the main process and then to the extrusion and
cooling line (labelled E&C in the figure). After a main process, a WIP needs curing time of about four hours. After
curing, the WIP needs to reenter a main process, which is available, for the next pass. The majority of products
undergo four passes.
In the main process passes having several passes, the compound recipes are converted to blends. After finishing
all required passes of the main process, the WIP goes on to the multiple passes to rubber sheets in the Final
machines.
In the main and final processes, there may be WIP batches with different quantities and types. The process
parameters are to be maintained, for example, with the temperature and product mix, semi-processed items from
previous batches, etc. The processing time of each operation may vary as well. In addition, the availability of the
machines after a specific set of time and distribution of the process where different passes are required among the
main and final machines are to be identified and monitored.
 Sriram Velumani and He Tang / Procedia Manufacturing 10 (2017) 100 – 111 103

1.3. Analysis Objectives

This study focus is on operation status, bottleneck, and possible improvement. The batching of product mix by
scheduling is currently experience based to govern the effect of batching on production while adding subsequent
batches for processing. The impacts of the availability of the machines and the workforce are less sufficient for
operations management. Advertently, the combinatorial effects of efficiency by the factors, including cycle time,
time between failure and time to repair, are also needed to be identified with help of simulation.
Considering various setup times, production volume, inventory, and quality issues affecting the outcome of the
products, the simulation analysis can visualize when the product batches are divided among machines for the factors
due to process control measure and these batches get mixed once they are in the system.
The simulation study is to examine the current setup. All machines arranged in series but the process may be
carried out by entering the same or different machine for the subsequent process pass, as the process requires
multiple re-entries.
The simulation study is also to find the bottleneck locations and the buffer size in the processes through efficient
routing. In other words, the study is to identify the bottlenecks and study the spots where the buffers should be
introduced or regulated.
Based on the simulation results, the study can propose a new process with a self-adjusting variable pull out
capacity, corresponding to the type of rubber sheets being rolled out by the machines. The machines can have less
pull out time and maintain the same quality. The simulation can also deal with ‘what if’ scenarios for further
improvements. In addition, the simulation model can be used to check the sensitivity of different machine layouts
with a new machine being included and analyzed in terms of queue size and waiting time.

2. Methodology

A comprehensive DES model is constructed. In the model, the throughput rate, availability, reliability, quality
(composite of all upstream and downstream) are considered. The quality of the units produced depends on the
process quality involving the raw material quality, process operating parameters, and influential factors such as
process temperature and other process variation, which are accounted in the effective process time of the mixer
machines.
The waiting time due to blockage or inventory level are to be monitored amidst factors like reliability and
availability of machine due to repair and unscheduled breakdown. The availability of the machine, movement of
WIP goods using manual operations and process constraints needs also to be considered.

2.1. Modeling Batch Process

The DES model considers the following the existing process setup and variables of the arrangement of the
machines. Based on the process flow, the overall process modeling is established, as shown in Figure 2 below. The
details of Main machines and Final machines are show in Figure 3.

Fig. 2. DES model of the manufacturing system

104 Sriram Velumani and He Tang / Procedia Manufacturing 10 (2017) 100 – 111

After finishing a process path, the products are palletized and taken from a machine to the next available machine
as seen in the production line model. The flow of the product depends up on the machine utilization in downstream
processes and the number of passes.

Fig. 3. Main machine unit and Final machine unit in simulation model

The input for the model are start and supply points, and are taken as fixed in time distribution as no significant
difference was measured. The reliability is taken as 95% in most of the machines based on the current situations.
The MTBF and MTTR for all main machines are between 10,500 to 12,000 batches and 120 minutes, respectively.
The values are same for all machines as they are identical and the Erlang distribution values are taken (k = 2 and 4)
for each machine as they were realistic [14]. The final process machines set to MTBF and MTTR 20,000 batches
and 200 minutes (including scheduled preventive maintenance), respectively.
A sample from every batch is collected once out of the machine and need to be checked and tested in the lab. The
test takes 20 – 25 minutes and the data is fed to the production line for go and no go.
The state of utilization of the main machines, such as Main 1 (M1), etc., is shown in the Figure 4. The machine is
not utilized to its full capacity due to process constraints and the raw material mixture specification changes of every
batch to be produced when compared to the standard values. Hence currently only 60% to 70 % of full capacity is
utilized (full capacity is 2,000 batches in main mixers and 1,500 in final mixers).

Fig. 4. Utilization of the main and final machines

 Sriram Velumani and He Tang / Procedia Manufacturing 10 (2017) 100 – 111 105

2.2. DES Modeling Considerations

The production throughput, or production rate (PR), is based on the average number of parts produced by the last
machine, which is the final mixer. The ratio of actual jobs per hour to the max jobs can be produced per hour taken
as 80 to 85%. In other studies, there is discussion on the criteria where the certain level of PR and line efficiency can
take considerable level of buffering and bottlenecks [15, 16, 17]. The modeling of this case as the routing for the
buffer from one machine to other is done for minimum wait time after four hours of curing. The visual logic code
and collection points in the form of work activity and queue are added in the DES model to perform the function. In
the present case, 85 units as the minimum level of buffer is taken. The buffer level can be identified from the
simulation model of how many units and ascertain the number of pallets and trolleys required for movement. The
exponential distribution applied with value near to ten machines have been discussed in [4, 18], which can also help
decide based on the line efficiency and the machine efficiency.
The process of rubber sheets has multiple (normally four) passes through the Main machines, which have buffer
stations, for example 4Hr M1 in Figure 3. The capacity of the buffer stations needs to be analyzed based on the
existing capacity and production rate, uptime, downtime, setup time, MTBF, and MTBR.
In addition to the Main machines, the buffer capacity of the Final machines F1, F2, and F3 should be given the
same consideration. The axiomatic way of implementing preventive maintenance as for improving reliability of the
machine is underway in the industry. To consider maintenance schedules and the efficiency of the machine, the
values are updated in the machines to model the real-time scenario. The model has built in functions such as stop
while maintenance for all machines which shows the semi-finished material routing through other machines. This
will enable to find alternate routes and operational bottlenecks. It is a quick procedure to understand and check
which other routing can work. The simulation for the reentrant lines, with serial combination along with reentrants
or repasses are not widely seen in [3, 19]. In this model with reentrant lines, the process timing, total batch time, or
time between maintenance (MTBF), workers’ technical skills are considered parameters in the model. The break
down time in the model is also considered for a what-if condition and rerouting is observed based on the earliest due
date.
The effective time for processing (ETP) can be considered as a method for making inputs to the machines with
comparatively fewer errors and reducing the process modeling complexity and is taken for all individual processes
for one complete batch (a type of WIP). This is considered while making the batch timings given in Table 1. It is
also cross checked with consideration from the employee’s observation and experience in the factory.

Table 1. Process schedule with typical batch items and processing information

Product type 1 2 3 4 5
Process
Occurrence
Time (min.)
No. of batches 200 300 85 120 450

9.5 First pass 1900 2850 807.5 1140 N/A 80%

8.5 First pass 168 2550 722.5 1020 N/A 10%

8 First pass 1100 2400 680 960 N/A 3%

10 First pass 120 3000 4250 1200 4500 90%

7.15 Regular Passes 720 2145 3040 858 3217.5 30%

8 Regular passes 730 2400 3400 960 N/A 10%

8.25 Regular passes 880 2475 3506 990 N/A 2%

9 Regular passes N/A 2700 3825 N/A N/A 5%

106 Sriram Velumani and He Tang / Procedia Manufacturing 10 (2017) 100 – 111

3. Results and Discussion

3.1. Simulation Results

The DES model is developed and conducted with the above factors and variables, such as run times and process
variations based on the observation on the floor. The data in Table 2 show the simulation results compared with
actual data, after a warm up time of 1,000 hours. The conditions are 180 batches per day (average per machine) in
the one main unit (M1 to M5), consisting of a mixer, extruder, and conveyor, and 325 batches per day (average per
machine) in the one final stage unit (F1, F2, and F3).

Table 2. Processing time for first passes and regulars compared with simulation

Process Product type Actual Simulation

Time (min.) No. of batches 200 85 200 85

8.5 First pass 1700 722.5 N/A 731

8 First pass 1600 680 N/A 692

8 Regular passes (3 passes) 1600 3400 865 3410

8.25 Regular passes (2 passes) N/A 3506 N/A 3511

Table 3. Simulation results based on the current situation

Machine Blockage time % Changeover time %

Main 1 7 1.04

Main 2 8 0.8

Main 3 12 0.6

Final 1 11 0.6

Final 2 7.2 0.7

Main 4 9 0.75

Main 5 10 0.6

Final 3 11 1.2

Table 3 shows the blockage and change over that occur for one shift. The simulation model just allocates the
routing based on the next available machine in the similar fashion done in the factory. Table 4 shows observed data
and the simulation results for the existing setup.

Table 4. Comparison between observed and simulated results

Utilization/Downtime of
Buffer fill (%) A) WIP (units) B) WIP (units) C)
Main machines (%)

Observed values 0.73/0.27 75% 255 120

Simulation results 0.78/0.22 72% 272 145

 Sriram Velumani and He Tang / Procedia Manufacturing 10 (2017) 100 – 111 107

In the table, the buffer status is the average of each shift (three shifts a day). A) indicates the buffer of the
current machine when it is processing with maximum output efficiency. B) denotes the number of units as WIP to be
processed but not ready in the main machines. C) denotes number of units as WIP to be processed but not ready to
be processed in the Final machine blocks.

3.2. Possible Improvement Measures

The opportunities to improve the production throughput based on the existing situation can be identified based on
the simulation. First, the pull-out efficiency of the mill in the machine units can be increased by adding hydraulic
and pneumatic circuits. This reduces the machine cycle time by 15 seconds for every batch. This is done as the
thickness of the sheet for every batch varies and it becomes slow making more lead-time. This engineering
suggestion was proposed and recommended by an engineer to take it forward.
Next, the allocation of the batches in the pallets consequently to other machine areas can be moved instead of
allowing them wait for the same machine. The conveyance from one machine to the other is around 85 feet and can
be done by the manual movement during idle time. This simulation study suggests the availability of the space to the
user. The third opportunity is to use the dispatching rule earliest due date and the jobs based on similar process time
and earliest due date. This is considered in the model for reducing any quality discrepancy.
Implementing the above suggestions, the simulated model shows that nominal batch size has no further effect of
processes as the sheets waiting (curing) for four hours. The finishing rate of jobs per hour for the overall process is
increased by two batches (jobs per hour). The buffer level in Table 4 indicates that the utilization of the machine is
higher. The higher utilization also indicates bottleneck occurrence downstream in the extruders, hence good
regulation can be considered and a steady buffer level (WIP) in the common inventory floor.
Decision making based on these values for preventive maintenance and balancing of time especially allocating
any additional task to other machines due to unavailability can be done [20]. There is a special case considered in
the model where the unavailability of the Main machines for any immediate job order requirement and testing
phases for small volume can be allocated to the Final machine for production as the main machines cannot be
rescheduled due to variation in process parameters. The DES model can be assigned with schedule for easy decision
making and planning in more units.

Table 5. Time (minutes) per batch output after the improvement in the model

Type 1 Type 2 Type 3 Type 4 Type 5

A1 B1 A2 B2 A3 B3 A4 B4 A5 B5

200 units 300 units 85 units 120 units 450 units

5 passes 2.45 2.40 2.50 2.42 2.55 3 2.45 2.45 2.40 2.50

6 passes 2.55 2.49 3.00 2.49 3.0 2.50 2.30 2.28 2.40 2.50

In the above Table 5, A – Actual time/batch to be palletized; B – simulation output / batch palletized. The regular
passes = Main machine passes + Final machine pass/passes. The 5 and 6 passes are about 80% and 10%,
respectively.
The process and product rolled out is always continuous, so indicating the time per batch is more applicable than
indicating the cycle time of the machines and some of the negligible cycle time for rarely occurring batches are not
considered. The time taken for each batch for the most-occurring batch size and cycle time is indicated in Table 5.
The logic instructs the process to take the next available machine. The model also suggests a route for the pallet. The
widely-used batch size and time utilized is shown here. Other batch sizes with rare occurrence are not considered as
the cycle time for them is even smaller and the number of passes required are typically 10 and 12. The very first
pass (requires blending all recipes) for all such requirements takes 5 minutes of machine cycle time.
108 Sriram Velumani and He Tang / Procedia Manufacturing 10 (2017) 100 – 111

As mentioned the routing of the batches with minimum wait time after four hours of wait time is applied in the
simulation. In Table 6 lists the max and average size of WIP of Main machines. The WIP status indicate the
performance level of every Main machine during operations.

Table 6: Buffer status from simulation

Main Machine Buffer WIP (average) WIP (max)

1 92.75 300

2 105 400

3 108.56 300

4 158.06 400

5 175 400

The routing logic for minimum change over (movement and setup time from pallet to next immediately available
machine) already instructed in the model. This helps the batching out of the machine for first to be serve or special
requirement like urgent orders other than the earliest due date specified in the model.
For the proposal of connecting the common inventory floor of all the machines, the simulation has been run
considering the parameters of directly linking the available pallets and specifying the change over time of 1.75
minute. This time is considered within the waiting time of four hours in the model. However, after the modification
is considered through simulation (all machines are connected seamlessly) the time for change over is analogous to
1.25 minutes between Area 1 and Area 2 or reduced by 30 seconds. An increase in production is seen by involving
Area 2 and a good load distribution among machines.

3.3. Discussion

It is noted the longest moving path is avoided unless there are more than two pallets to be processed immediately
or from the simulation. It can be learned where in the inventory zone to store the pallet and which might be closest
to the next available machine. Some unwanted movements can be avoided. The method of allocating and
reallocating is assigned by simulation by giving the equivalent number of passes (jobs) to all Main and Final
machines. This will enable a balanced maintenance for all machines.
In the simulation, the buffer size is determined as 84 pallets per machine for a given batch size to be processed in
the machines of this re-entrant configuration. The actual production site is executing the chain as fast as it can and
pull in more “made-to-order” batches without delay using Kanban procedures. The typical batch size is listed in
Table 1.
Note the sheets which arrive out of the Main machine and the Final machine pass through a cooling conveyor and
then through a vertical conveyor to get palletized. The cooling conveyor is in order to reduce the long conveyor line
of 85 feet due to space constraint and process time.
The work-in process is considered as total conveyance time plus the wait time. WIP total = cooling conveyance
time + conveyance time + wait time 4 hours + feeding time for further passes (set up time). Setup time is considered
between the roller mills both in the Main machines and the Final machines. Transfer time between machines is
considered as the feed time + change over time.

Table 7. The WIP in the process for the buffer (pallets) from simulation

Conditions Utilization/transfer time (%) Buffer fill (%)

A) Existing setup 0.91/0.09 86

B) Connected setup 0.95/0.05 88

 Sriram Velumani and He Tang / Procedia Manufacturing 10 (2017) 100 – 111 109

In the Table 7, A) is the existing situation without being connected and no dedicated setup. In addition, (B) is all
Main and Final machine units are connected for easy movement of WIPs.
Worker training can be set using the simulation model with trial runs and will help us in finding the efficiency of
the line when different workers with skill are implemented.
The throughput for overall batch size is increased by 2% in implementing the varying pullout mechanism which
helps in reducing the time and effectively improve the pull-out procedure for feeding into the roll mills. The time
taken here is reduced, which is directly included as cycle time reduction of 6 seconds in the model properties of the
main mixture machines. The impact in the downstream is found to be very less causing few bottlenecks in the mill
rollers.
Dispatching the equal distribution of the types and batches among machines always need thorough monitoring
and scheduling problems. The model output shows based on the logic there is nearly an uniform allocation of
batches according to the number of passes is made this regulates the utilization rate among the machines each time
the new batch orders are taken, refer to Table 8. This enables the planning for more order handling and less tardiness
among machine output from each unit. The current rate of processing is set for around 2.75 to 3 minutes for efficient
make span of 180 batches per day for main machines which can be reached through continuous prediction using
DES. The maintenance rate of the machines can be assessed without much difficulty for identifying the machine
time slots in seasons where the orders are different than expected with more number of passes and quality
requirements.

Table 8. The batch completion per shift and average utilization rate

Number of batches
Machine Utilization %
per shift
Main 1 83 176

Main 2 79 170

Main 3 86 182

Final 1 80 364

Final 2 82 371

Main 4 85 179

Main 5 80 173

Final 3 80 356

4. Conclusions

This study addresses a moderate complex batch processing with considerations of product and process variables
based on simulation. A comprehensive DES model is built for the analysis of production throughput of a serial re-
entrant line with a batch production process and product/process variations. In the simulation, main process and
production variables such as bottlenecks, utilization, and process time for same and different batches are compared
with the priority on due date. The dispatching based on the earliest due date is used. The simulation logic is used to
check the batch allocation for less tardiness in all machines and provide expected time. The evolving simulation
outputs help identify how the production proceeds in a specific batch size as well as production bottlenecks through
priority with routing logics. The analysis results have a good agreement with floor observation, which also provides
a good understanding of the operations status of machines and buffers. By including the pullout based changes and
integrating the inventory of Area 1 and Area 2, the requirement of machines is found form the simulation. This is
then checked with the reduction in process time and the relationship among the expected reductions of bottleneck,
tardiness, and production efficiency. The integration helps in reducing the unnecessary waiting time and easy
110 Sriram Velumani and He Tang / Procedia Manufacturing 10 (2017) 100 – 111

movement with reduction by 30 seconds. All the above changes are focused on reducing bottleneck, tardiness, and
production efficiency by posing those improvements in simulation.
The options are given for machine maintenance or breakdown that can be used for finding an alternate routing of
the batches while the next immediate machine is unavailable. Based on the simulation, therefore, some improvement
proposals are evaluated for the improved operation throughput, such as the overall process time. The proposal for
installing a new machine can be ascertained by the simulation for about a week and identifying through dispatching.
The buffer level and status are instantly updated and can be checked at a preferable time of the shift.
The DES modeling study is on the main process of tire manufacturing and can be extended to the entire plant for
the same purpose of understanding for each batch size moving with dispatching rules. The simulation model can be
updated with data systems or SCADA for easy update of machine conditions and process parameters for enabling
more logic and machine conditions relevant to batch processing of different types.

Acknowledgements

The authors are sincerely grateful for the support from the EMU Provost’s Faculty Award and EMU graduate
assistantship. The authors also thank Mr. Brian Harrington for his help on the Simul 8 modeling.

References

[1] Anil Khurana, Operations, Supply Chains & Logistics Managing Complex Production Processes, (1999), Available from:
http://sloanreview.mit.edu/article/managing-complex-production-processes/
[2] Yang Liu, Jingshan Li, and Shu-Yin Chiang, Re-entrant Lines with Unreliable Asynchronous Machines and Finite Buffers: Performance
Approximation and Bottleneck Identification, Int. Journal of Production research, 50:4, 977-990, DOI:10.1080/00207543.2010.550639.
[3] Jingshan Li and Semyon M. Meerkov, Production Systems Engineering, 1st ed. Springer, 2009
[4] Christoph Laroque et al, Increase of Robustness on Pre-optimized Production Plans Through Simulation-based Analysis and Evaluation,
SIMUL 2011: The Third International Conference on Advances in System Simulation
[5] J. W. Fowler and O. Rose, Grand Challenges in Modeling and Simulation of Complex Manufacturing Systems, SIMULATION, vol. 80, no.
9, pp. 469–476, Sep. 2004.
[6] Ningxuan Kang, Li Zheng, and Jingshan Li, Analysis of Multi-product Manufacturing Systems with Arbitrary Processing Times,
International Journal of Production Research Vol 53. (2015), No 3 pp 983 to 1011
[7] Filippo De Carlo, Maria Antonietta Arleo, and Mario Tucci, OEE Evaluation of a Paced Assembly Line Through Different Calculation and
Simulation Methods: A Case Study in the Pharmaceutical Environment, International Journal of Engineering Business Management Special
Issue: Innovations in Pharmaceutical Industry, August 2014.
[8] I. Carson and S. John, Introduction to Modeling and Simulation, Proceedings of the 36th Conference on Winter Simulation, pp. 9–16, 2004.
[9] Can Cogun and Mehmet Savsar, Performance Evaluation of a Flexible Manufacturing Cell (FMC) by Computer Simulation, Modeling,
Measurement and Control B 62(1-2):31-43, Jan 1996
[10] Fenglan He, Dieter Armbruster, Michael Herty, and Ming Dong, Feedback Control for Priority Rules in Re-entrant Semiconductor
Manufacturing, Applied Mathematical Modeling 39. (2015) pp 4655–4664
[11] R. Singh, D. B. Shah, A. M. Gohil, and M. H. Shah, Overall Equipment Effectiveness (OEE) Calculation — Automation through Hardware
& Software Development, Procedia Eng., vol. 51, pp. 579–584, 2013.
[12] Christoph Roser, Masaru Nakano, and Minoru Tanaka, A Practical Bottleneck Detection Method, DOI: 10.1109/WSC.2001.977398, Winter,
2001. Proceedings of the Winter Simulation Conference, Vol. 2
[13] Adrián Aguirre, Enrique Müller, Sebastián Seffino, and Carlos Mendez, Applying A Simulation-Based Tool to Productivity Management in
An Automotive-Parts Industry, Proceedings of the 2008 Winter Simulation Conference
[14] Emre enginarlar1, Jingshan Li and Semyon M. Meerkov, How Lean can Lean Buffers Be? IIE Transactions (2005) 37, 333–342 DOI:
10.1080/07408170590916878.
[15] Pankaj Sharma, Ajai Jain, Performance Analysis of Dispatching Rules in a Stochastic Dynamic Job Shop Manufacturing System with
Sequence-Dependent Setup Times: Simulation Approach, CIRP Journal of Manufacturing Science and Technology, 2015.
[16] Maheshwaran Gopalakrishnan, Anders Skoogh, and Christoph Laroque, Simulation-Based Planning of Maintenance Activities in The
Automotive Industry, Proceedings of the 2013 Winter Simulation Conference.
[17] Mukund Subramaniyan, Anders Skoogh, Maheshwaran Gopalakrishnan, and Atieh Hanna, Real-time Data-driven Average Active Period
Method for Bottleneck Detection, International Journal of Design & Nature and Ecodynamics 11(3):428-437, July 2016
[18] Steve H. Wang, Shu-Ping Chang, Paula Williams, Benjamin Koo, and Yan-Rui Qu, Using Balanced Scorecard for Sustainable Design-
centered Manufacturing, 43rd North American Manufacturing Research Conference, NAMRC 43, 8-12 June 2015
 Sriram Velumani and He Tang / Procedia Manufacturing 10 (2017) 100 – 111 111

[19] Boris Ljubenkov, Goran Dukic, and Marrino Kuzmanic, Simulation Methods in Shipbuilding Process Design, Journal of Mechanical
Engineering 54 (2008) 2, 131-139 UDK-UDC 629.5.081.
[20] Brian Harrington, The Balancing Act: An Example of Line Balancing, Available from https://www.simul8.com/resources/an-example-of-
line-balancing-with-simulation.pdf, 2015.