A worker assignment for machine cluster in the
manufacturing cell was discussed in this paper. The
objective of this paper is two folds. First, it aims to
demonstrate a calculation of the maximum number of
machines which one worker can optimally operate without
an idle time on the machine. Then, the simulation software
program is utilized to evaluate various proposed
improvement alternatives based on maximum machine
efficiency criteria. In this study one of the hard disk drive
manufacturers in Thailand was explored. Previously, one
worker was assigned to service eight semi-automated
machines in the cluster. The main tasks of the worker were
loading, checking and recording. Due to the fact that
workload of the worker and the machine cycle is not
balanced so the utilization of the worker cannot be
maximized. Therefore, the optimum number of machines
which one worker can ideally service is calculated. It is
found that one worker can optimally serve up to eleven
machines. However, other alternatives are also explored. As
a result, by systematic work elimination, one worker can
service up to fourteen machines per cycle with 96.09%
machine utilization.
Keywords – Worker assignments, machine cluster, single
manufacturing cell
I. INTRODUCTION
The single automated cells constitute the most
common system in the hard disk drive manufacturing
system industry. This cell operates independently of other
workstations in the factory [1]. It composed of a fully
automated machine which can be of unattended operation
for a longer period of time. This type of machine usually
equips with an automatic pallet change (APC) and
automatic tool change (ATC). The work-parts are kept at
the part storage subsystem and will be automatically
transferred and positioned to the machine. Due to the fact
that this machine does not require continuous attention
from an operator during its semi-automatic machine
cycle, the operator can be assigned more than one
machine to operate. A worker assignment task normally
is handled heuristically particularly at the cell
implementation level. Matching of worker experiences to
a production requirement is the most common practices.
This can lead to an unbalance workload, non utilization of
machine, and low efficiency. The company may not be
able to maximize utilization of workers and machine [2].
The number of research has been done in this area.
Most of the research papers employ mathematical models
and linear programming technique to attain the solution.
Examples of these papers will be briefly discussed here.
A simultaneous consideration of machine and worker
based on the worker skill through a multi-objective
mathematical model was proposed in 1993 [3]. The
optimal manpower assignment in manufacturing cells was
attempted using mixed integer programming and then
integer programming in 1996 [4]. It is a two-step
hierarchical method. Likewise, the integer programming
model was applied to assign workers to the manufacturing
cells and later another integer programming model was
utilized to schedule appropriate training program for these
workers in 1997 [5]. Similarly, another mixed integer
programming model was presented to verify that a
manufacturing cell was improved when worker skill is
taken into account for worker assignment and training
purpose in 2002 [6]. Another application of mixed
integer programming was used to assign workers to the
cell with an aim to minimize total intra-cell workforce
transfers in 2005 [7]. A goal programming model was
also implemented to design a cell and then assign workers
to these cells [8]. Again, an integer programming model
was designed to select workers for cross-training in the
cell [9]. A simulation model was also used to analyze
factors influencing the flexibility of cellular
manufacturing system [10]. It was found that cross trained
operators play an important role in the flexibility of the
cell. The similar result was obtained by another research
paper using decision rules together with simulation model
[11]. A Markov decision process was used to analyze the
performance of the manufacturing cells and revealed that
capacity balance and variability buffering can improve
performance of the cells [12]. Utilizing workload balance
as the main factor for assigning workers in the cell was
another application by the simulation model [13].
In terms of quantitative approach, there is an attempt
to decide the most efficient number of workers and
measurement method in the manufacturing cell. This
accomplished by data envelopment analysis in decision
model [14]. Recently, there is attempt to apply artificial
intelligent in the worker assignment problem. An
approach based on artificial neural network was proposed
[15].
It can be seen that most worker assignment problems
were accomplished using a mathematical model and a
simulation technique. Only a few apply artificial
intelligence together with heuristic rules. In this paper, a
quantitative approach is presented. Then a simulation
A Worker Assignment for Machine Cluster in the Manufacturing Cell
Suksan Prombanpong, Waraporn Seenpipat
Institute for Scientific and Technological Research and Services,
Department of Industrial Engineering, King Mongkut’s University of Technology Thonburi, Bangkok, Thailand
(suksan.pro@gmail.com)
978-1-4577-0739-1/11/$26.00 ©2011 IEEE
352
model is utilized to compare performance of all
alternatives.
II. METHODOLOGY
1) Problem Description: In order to produce a hard
disk drive, many precision parts must be produced and
assembled. One of the most important parts is the motor
spindle shaft. This part must be machined by an automatic
machine which requires less effort from an operator.
Therefore, an operator is assigned to perform 3 different
tasks. The first one is a direct service to the machine such
as loading/unloading the work-part bin, inspection the
finished goods, correction minor machine problem and
reset alarm warning. The second task concerns an off-line
work-part preparation, counting and transporting the
finished work-parts. The last assignment is recording data
i.e. tool change and adjustment data, production output
and inspection data. It can be recognized that the operator
does not require attention to the machine at all time.
Consequently, one operator is able to service more than
one machine at a time. The total number of machines
which one operator can optimally handle without a
machine idle time is considered a machine cluster
problem.
The current manufacturing cell consists of 188 CNC
machines working in three shifts. Each shift needs 23
operators to attend these machines. Therefore, each
operator is currently assigned to handle eight machines at
a time. The current number of machines which one
operator can consecutively handle was not derived from
any calculation but merely from judgment. As a result, an
operator idle time at the end of each cycle can be
expected. Furthermore, the production line plans to
increase more machines in the cell. The management does
not only want to increase any more operators but also
expect to double a number of machines assigned to each
of the operator. Again, this number is from imagination.
Thus, a feasibility study must be performed.
2) Problem Solving: As mentioned earlier, the
number of machines assigned to one operator is not
optimum. Thus, the optimum number of machine must be
calculated. The solution can be obtained from (1). In this
equation, the nominator is a total time that a machine
needs to complete in a cycle. Note that this is a
combination of the machine cycle and the operator service
time i.e. unloading and loading time at the end of the
cycle. The denominator is a total time that an operator
must perform on the machine in addition to walking time
or repositioning time among the machine cluster.
N= Maximum integer ≤ os r
mc os
T T
T T
(1)
N= A group of machines which can be serviced by one
operator and it must be an integer; Tmc = machine cycle
time (min); Tos = operator service time to a machine
(min); Tr = repositioning time or operator walking time
among machine cluster (min). The data of the above
variables are collected and shown in Table I.
TABLE I
THE ESSENTIAL DATA FOR CALCULATION
Description Time(min)
Machine Cycle Time
Service Time
Repositioning Time
88.67
8.71
0.09
Thus, substitute Tmc= 88.67, Tos= 8.71 and Tr = 0.09 in
(1).
N ≤
8.71 0.09
88.67 8.71
Thus
N ≤ 11
As a result the optimal number of machines which one
worker can serve is equal to eleven machines. However,
the management prefers to explore higher number of
machines. Thus, an elimination of some activities together
with rearrangement in order to reduce operator service
time is inevitable. Therefore, a number of machines
varied from 12 to 14 are investigated. A recalculation of
Tos in (1) based on different number of N must be
performed to measure amount of work needed to be
eliminated. The result of calculation is shown in Table II.
It is obvious that if one worker must service more
machines in the cell, the amount of worker service time
must be reduced for compensation. As an example, for
12-machine, the operator service time will be 7.96
minutes and at least 45 seconds of work must be reduced.
Therefore, some unimportant tasks should be eliminated
equal to or greater than a reduction time. Typically, nonvalue
added works should be prioritized and
consecutively eliminated until no further works can be
logically discarded. A work value analysis is shown in
Table III. A list of works including their operation times
are shown in Table IV. Then, all of the alternatives are
simulated by a simulation software program. The
selection criterion is maximum machine utilization and
will be described in details next.
TABLE II
THE CALCULATED AMOUNT OF OPERATOR SERVICE TIME
No. of Machines Tos (min) Reduction Time (sec)
12-Machine
13-Machine
14-Machine
7.96
7.26
6.72
45.0
85.2
119.4
353
TABLE III
THE WORK VALUE ANALYSIS
Activity VA NVA Necessary
Check w