AN EXPERIMENTAL STUDY ON ASSEMBLY WORKSTATION

CONSIDERING ERGONOMICALLY ISSUES

Ibrahim H. Garbie

Department of Mechanical and Industrial Engineering, Sultan Qaboos University
P.O.Box 33, Al-Khoud 123, Muscat, Sultanate of Oman, Email: [email protected]

ABSTRACT
This paper describes the results of an experimental study conducted to investigate the effects of assembly of a
product on operator performance. A fully adjustable ergonomically designed assembly workstation (smart
workstation) was used for the experiment. Ten college students (five boys group and five girls group) randomly
assigned into three experimental conditions (table adjustable, chair adjustable, and gender) performed the
assembly task. Performances of the participants assembling a product are: operator production rate representing
in how many assembly products per unit time (units/hour). The regression model to measure the operator
performance was built based on the experimental work using Minitab Statistical Software package. The results
shows that female are more productive than male.

Keywords: Ergonomic design, flexible workstation, experimental design

1. INTRODUCTION
Ergonomics is concerned with making the workplace
as efficient, safe and comfortable as possible.
Effective application of ergonomics in work system
design can achieve a balance between worker
characteristics and task demands. This can enhance
operator productivity, provide worker safety and
physical and mental well-being and job satisfaction.
Many research studies have shown positive effects of
applying ergonomic principles in workplace design,
machine and tool design, environment and facilities
design [1-9].
Research studies in ergonomics have also produced
data and guidelines for industrial applications. The
features of ergonomic design of machines,
workstations, and facilities are well known [10-17].
However, there is still a low level of acceptance and
limited application in industries, especially in
developing countries. The main concern of work
system design is usually the improvement of
machines and tools. Inadequate or no consideration is
given to the work system design as a whole.
Therefore, poorly designed work systems are a
common place in industry [4 and 11]. Neglect of
ergonomic principles brings inefficiency and pain to
the workforce. An ergonomically deficient workplace
can cause physical and emotional stress, low
productivity and poor quality of work [18-19].
Workstation should be laid out such that it minimizes
the working area so that while carrying out the
operations the worker could use shorter motions and
expend less energy and thus reduce fatigue. Das and
Grady [12] reviewed the concept of workspace design
and the application of anthropometric data. It
indicated that an adjustable chair and a workbench of
standard size were highly desirable at the workplace.
However, the standard height of the workbench could
not be defined without the anthropometric data of the
user population. Many of the user population do not
have anthropometric data. It is therefore, desirable
also to have the worktable adjustable [20-21].
A study by Yeow [22] concentrated on improving
productivity as well as health and safety of workers in
a printed circuit assembly (PCA) factory. The
improvement involved the use of an ergonomically
designed workstation with other ergonomic
intervention such as clear segregation of tested and
untested boards to prevent mix-up and retraining of
operators by more qualified trainers. This had
resulted in an improvement in quality and
productivity of the workers, reduction in rejection
rate as well as an increase in the revenue. The use of
an ergonomically designed workstation and better
structured processes along with other features, such
as improved lighting, shelves and containers for parts
and display boards, had helped and solved the
problems of assembly processes at a German
company [23]. The objective of this research was to
study the productivity of operator by assembly a
product on the smart workstation for a repetitive
industrial assembly task taken into consideration
table, chair adjustable and type of gender.

2. METHODOLOGY
The experimental study was conducted in the
Ergonomics Lab of the Department following a sound
Proceedings of the 41st International Conference on Computers & Industrial Engineering
275

methodology. Details of the study elements are
described in the following sections.

2.1 The Task
The selected task was an assembly task from a local
electrical company, assembly of fused electrical
switch that consisted of eight parts. Usually,
simulated tasks are chosen for research purposes that
do not represent real life industrial tasks. Manual
assembly of switches is a common task in electrical
industry. The selected task was a highly repetitive
task and it was performed on workstations that were
not designed ergonomically. Also, the task method
was not designed following ergonomic principles.
The assembly task involved picking up the switch
base and cover from the bins, assembling all the
inside parts in the base, putting the cover, tightening
the assembly using a screwdriver and placing it in the
outgoing bin. The steps of the assembly task were
modified in the new design considering motion study
and ergonomic principles. A jig was designed and
used in the performance of the task on the smart
workstation. A power screwdriver was used for
tightening the cover in the modified task method.

2.2. Participants
Ten college students (five boys group and five girls
group) participated in the experimental study on a
voluntary basis. The average age of the participants
was 21.5 yrs with a standard deviation of 1.11 yrs.
Mean stature was 1850 mm with a standard deviation
of 101 mm. This indicated a significant size
difference among the participants. The participants
had no prior experience on the assembly task. They
were given instructions on the assembly workstations
and task and trained for 15 minutes on the task, as
required based on their experimental conditions.
Fifteen minutes training was considered adequate as
the assembly task was not a complex task according
to the learning rate. Environmental condition (light,
temperature, humidity and noise) was comfortable
and kept constant. The participants wore light and
comfortable clothes.

2.3. The Experimental Study
Experiments were conducted using an ergonomically
designed smart assembly workstation. Details of the
ergonomically designed smart assembly workstation
were reported in [11].

2.3.1. The Smart Assembly Workstation
The smart assembly workstation was designed and
developed considering ergonomics in all aspects of
design and layout with full adjustability. The size of
the tabletop (work surface) was calculated based on
the mean reach of the user population with an
allowance. A special table frame was designed for the
vertical and angular movements of the tabletop using
small motors. The frame mechanism was designed for
precise movements of the tabletop. Push-button
switches were provided for the control of these
movements. Operators could adjust the tabletop to
their most preferred work heights. The table could be
used for sit, stand, and sit-stand assembly
workstations. Attachments were provided to the frame
for bins and tools holders for adjustments. A fully
adjustable ergonomic chair was provided to the
operators. Major features of the ergonomic adjustable
chair were: adjustable seat height by gas suction,
adjustable and titled back support, tilted seat pan
covered with porous and breath-able material,
removable and adjustable arm rests, footrest and a
foot ring.
An adjustable hydraulic footrest was provided for the
operators. The existing hand tools were replaced with
a power screwdriver that was supported by a balancer
in front of the operator. The workplace layout was
made according to the calculated normal and
maximum work areas. Squires method was adopted
in the calculation. The bins were laid out based on
this calculation and in a logical work sequence and a
systematic method. Figures (1-2) show the isometric
view and the schematic layout of the ergonomically
designed smart assembly workstation, respectively.
An improved work method following the assembly of
parts sequence was developed for the task
performance on this workstation. A jig was designed
for ease of holding the base of the switch.

2.3.2 Experimental Setup
The ergonomically designed smart workstation was
installed and set up in the Ergonomics Lab. Tables 1
shows the experimental conditions. Experiments were
conducted at random times but not in the same week
on both groups. Boys group was implementing their
tasks in different time than girls group. Three factors
are considered in the experimental work: table
adjustable; chair adjustable and gender. With respect
to table adjustable, there are five levels of
experimental to adjust the table taking into
consideration the ground floor as a reference point:
23.5, 27.5, 31.5, 35.5, and 39.5 cm. Also with respect
to chair adjustable, there are five levels: 18.5, 19.75,
21, 22.25 and 23.50 cm. Regarding the gender, the
conducted experimental includes two levels (male and
female). Each experimental was conducted twice
(number of replicates = 2) and the performance
measurements are recorded based on performance
measure: operator output (e.g., production rate).
Participants were given a demonstration and then
trained for 15 minutes the smart workstations and
methods before starting the experimental sessions.
Each participant had assembled electrical switches for
Proceedings of the 41st International Conference on Computers & Industrial Engineering
276

one hour duration under his experimental condition
randomly and the operator performance was recorded
in terms of number of switches assembled (units/hr),
operator satisfaction and operator health. A complete
factorial design for different levels of independent
variables is planned in 100 experimental (50 setups)
with two replicates for each response. Tables 2 is
used to display the observed data from the conducted
experimental.



Figure 1: Isometric view of the workstation [21]


All dimensions are in cm

Legend: 1. Pin 1 bin; 2. Pin 2 bin; 3. Connector
bin; 4. Screw bin; 5. Cover 2 bin; 6. Fuse bin; 7.
Pin 3 bin; 8. Cover 2 bin; 9. Outgoing bin; 10.
Power screwdriver; 11. Assembly area; 12. Fixture;
13. Assembly table; 14. Maximum reach; 15.
Normal area proposed by Squires.

Figure 2: Schematic layout of the workstation [21]

Table 1: Experimental conditions with different factors and levels
Factors Levels
Table adjustable (cm), (T) 23.5, 27.5, 31.5, 35.5, 39.5
Chair adjustable (cm), (C) 18.5, 19.75, 21.0, 22.25, 23.50
Gender (type), (G) Male (M), Female (F)

Table 2: Data from experimental conditions with Operator output (Units/hour)
Gender (G)
Male (M) Female (F)
Table adjustable (cm), (T) Table adjustable (cm), (T)
23.50 27.50 31.50 35.50 39.50 23.50 27.50 31.50 35.50 39.50



Chair
Adjustable
(cm), (C)
18.50 116
119
80
83
90
83
72
87
72
76
146
157
148
163
182
178
211
226
173
224
19.75 105
109
108
105
108
117
76*
87
65
94
131
132
146
119
189
208
169
137
137
154
21.00 98
72
94
87
98
98
98
87
76
54
134
138
152
184
135
112
170
103
187
193
22.25 83
65
76
105
108
108
105
80
87
101
115
109
179
124
127
143
160
154
154
175
23.50 69
58
69
80
83
83
83
137
98
98
165
143
158
198
223
297
165
158
156
171
(*) is the measured value of production rate (units per hour)
Table adjustable
Chair adjustable
Proceedings of the 41st International Conference on Computers & Industrial Engineering
277
3. RESULTS AND DISCUSSION
The operator performance data were summarized in
Table 2 and analyzed using Minitab Statistical
Software Package for analysis of variance (ANOVA)
and regression models for each performance measure
sequentially. The data presented in Table 3 are
analyzed with the analysis of variance (ANOVA)
technique. It seems from Table 3 that the main effects
of the three factors (T, C and G) and the interactions
effects (TC, TG, CG and TCG) are significant on
production rate based on p-values which are less than
0.05. For this reason, they have been included in a
regression model to build a mathematical formulation
between these factors and production rate. It can be
observed from Figure 3 that table adjustable with
third level (31.50 cm) is the highest on production
rate among all levels and there is no difference
between fourth and fifth levels. This means that it is
not needed to raise a table up to 39.50 cm (fifth
level). With respect to chair adjustable, the first level
(18.50 cm) and the fifth level (23.50 cm) are
representing the highest values on production rate and
there is no difference between third and fourth levels
in the chair adjustable. This means it does not matter
to raise a chair up to 21.00 cm (third level) or 22.25
cm (fourth level). Regarding gender, female are more
productive than male in the assembly stations. With
respect to interaction effect between table adjustable
and chair adjustable, it can be noticed from Figure 4
that third level of table adjustable (31.50 cm) with
fifth level of chair adjustable (23.50 cm) is
representing the highest value of production rate.
Regarding interaction effect of table with gender, it
can be observed from Figure 5 that table adjustable
with third level (31.50 cm) with female represents the
highest values in production rate among all levels.
Also, the interaction effect between chair adjustable
and gender is producing the highest value in
production rate with first level and/or fifth level of
chair adjustable with gender (female) (see Figure 6).
It is recommended hiring female in assembly
workstation especially which has adjustable chair and
table. It can be concluded from this analysis that third
level of table adjustable; fifth level of chair adjustable
with female will give high productivity of operator
performance.


Table 3: Analysis of variance for production rate
Source DF SS MS F P-value
______________________________________________________________________________
Table adjustable 4 6628.7 1657.2 5.91 0.001
Chair adjustable 4 5810.2 1452.5 5.18 0.001
Gender 1 129024.6 129024.6 460.08 0.000
Table adj*Chair adj 16 13595.3 849.7 3.03 0.001
Table adj*Gender 4 5199.7 1299.9 4.64 0.003
Chair adj*Gender 4 9461.4 2365.3 8.43 0.000
Table adj*Chair adj*Gender 16 22572.3 1410.8 5.03 0.000
Error 50 14022.0 280.4
Total 99 206314.2


5 4 3 2 1
160
140
120
100
5 4 3 2 1
2 1
160
140
120
100
Table
M
e
a
n
Chair
Gender
Main Effects Plot for Production Rate
Data Means


Figure 3: Main effects of table adjustable, chair
adjustable and gender on production rate



5 4 3 2 1
180
170
160
150
140
130
120
110
100
90
Chair
M
e
a
n
1
2
3
4
5
Table
Interaction Plot for Production Rate
Data Means


Figure 4: Interaction effect of table and chair on
production rate

Proceedings of the 41st International Conference on Computers & Industrial Engineering
278

2 1
180
160
140
120
100
80
Gender
M
e
a
n
1
2
3
4
5
Table
Interaction Plot for Production Rate
Data Means


Figure 5: Interaction effect of table and gender on
production rate


2 1
180
160
140
120
100
80
Gender
M
e
a
n
1
2
3
4
5
Chair
Interaction Plot for Production Rate
Data Means


Figure 6: Interaction effect of chair and gender on
production rate


The response function representing the production
rate (P) is expressed as the following equation (1):

) , , ( G C T f P (1)

Where:
P: the production rate (response) or yield
(units/hour),
T: table adjustable,
C: chair adjustable,
G: gender

A regression model is used to present the results of a
designed experiment in a quantitative form. The
second-order polynomial is capable of assuming a
wide variety of shapes and it is a very flexible
regression model [24-25]. The second order
polynomial (regression) equation is used to represent
the response (production rate) for K factors by using
the following equation (2):

2
3
1
3
1 ,
3
1
i
K
i
ii j i
K
j i
ij I
K
i
i o
X B X X B X B B P

(2)

Where:
o
B is the free term of the regression model,
) ,....., , (
2 1 K i
B B B B are the linear terms,
) ,....., , (
1 13 12 K ij
B B B B are the interaction terms,
) ,....., , (
22 11 KK ii
B B B B are the quadratic terms

The values of the coefficients of the polynomial of
equation (2) are calculated by the regression model.
The Minitab Statistical Software Package has also
been used to calculate the values of these coefficients.
The mathematical model as determined by above
analysis is given as the following equation (3) and it
is considered as a full initial regression model
representing the production rate (units/hour) of
assembly smart workstation.

2 2
36 . 4 27 . 3 66 . 0 81 . 8
31 . 1 4 . 43 7 . 31 7 . 5 3 . 72
C T CG TG
TC G C T P
Inital


(3)

The
2
G term (gender) has been removed from the
equation through the Minitab Statistical Software
Package because it has highly correlated with other
variables. Summary of initial full regression model
for production rate estimation is shown in Table 4. It
can be noticed from Table 4 that C, G,
2
T ,
2
C and
TG interaction were found to have significance on
production rate although C and
2
T have negative
effects but T, TC and CG have no significance effect
based on p-values (p < 0.05). Testing of significance
of regression model is evaluated through p-value
equals 0.00 less than 0.05 (95.00% confidence level)
although the determination of coefficient of initial
regression model (
2
R ) was 69.8 % and the
associated adjusted determination of coefficient (
2
R -
adj) was 67.2%.





Proceedings of the 41st International Conference on Computers & Industrial Engineering
279

Table 4: Summary of initial full regression model of production rate
_______________________________________________________________
Predictor Coefficient SE Coef T-test P-values
_______________________________________________________________
Constant 72.35 32.59 2.22 0.029
Table 5.66 11.73 0.48 0.631
Chair - 31.74 11.73 - 2.71 0.008
Gender 43.43 16.54 2.63 0.010
Table*Chair 1.310 1.308 1.00 0.319
Table*Gender 8.810 3.699 2.38 0.019
Chair*Gender 0.660 3.699 0.18 0.859
Table^2 -3.268 1.563 -2.09 0.039
Chair^2 4.364 1.563 2.79 0.006
_______________________________________________________________
S = 26.1560 R-Sq = 69.8% R-Sq(adj) = 67.2%
_______________________________________________________________
ANOVA for testing significance of initial regression model______
Source DF SS MS F P-value
Regression 8 144058 18007 26.32 0.000
Residual Error 91 62256 684
Total 99 206314
________________________________________________________________

When
2
R and
2
R -adj are not different dramatically,
there is a good chance that significant terms have
been included in the regression model [24-25]
although
2
R and
2
R -adj are not large enough.
However, as it has noted in Table 4 that a large value
of
2
R and
2
R -adj does not necessarily imply the
regression model is a good one and provide accurate
predictions of future observations.
2
R is a measure of
the amount of reduction in the variability of
production rate by using the regressor variables. It is
recommended to drop the insignificant terms (T, TC
and CG) in the initial full regression model to let it
more accurate, easy manipulate and consistency [26].
These data are presented in Table 5 and are
considered a modified regression model. The new
modified mathematical model of production rate
determined by the modified regression model is given
as the following Equation (4). It can be observed
from Table 5 that all independent variables (C, G,
2
T ,
2
C and TG interaction) were found to have
significance on the production rate with little bit
changes in
2
R and
2
R -adj although C and
2
T still
have negative effects on production rate. The final
summary of the experimental work is presented in
Table 6.

2 2
36 . 4 12 . 2 4 . 10 6 . 40 8 . 26 7 . 73 C T TG G C P
Modified

(4)

Table 5: Summary of modified regression model of production rate
_____________________________________________________
Predictor Coefficient SE Coef T P-values
_____________________________________________________________
Constant 73.69 17.26 4.27 0.000
Chair -26.816 9.497 - 2.82 0.006
Gender 40.58 10.86 3.74 0.000
Table*Gender 10.421 3.178 3.28 0.001
Table^2 -2.1174 0.8218 - 2.58 0.012
Chair^2 4.364 1.553 2.81 0.006
___________________________________________________________
S = 25.9867 R-Sq = 69.2% R-Sq(adj) = 67.6%
____________________________________________________________
ANOVA for testing significance of modified regression model_
Source DF SS MS F P-value
Regression 5 142835 28567 42.30 0.000
Residual Error 94 63479 675
Total 99 206314
____________________________________________________



Proceedings of the 41st International Conference on Computers & Industrial Engineering
280


Table 6: Summary of Experimental work
Performance measure Table adjustable
(T)
Chair adjustable
(C)
Gender (G)
Male Female
Productivity
(P)
Factors 3
rd
level 5
th
level Significant
Model
2 2
36 . 4 12 . 2 4 . 10 6 . 40 8 . 26 7 . 73 C T TG G C P




4. CONCLUSIONS AND
RECOMMENDATION

The following conclusions were drawn from this
experimental study:
1. Operators performance with regard to
productivity with the ergonomically smart
assembly workstation condition is studied
and investigated.
2. The fully adjustable ergonomically designed
smart assembly workstation was preferred by
the operators and they adjusted and
organized the workstation to their comfort.
3. Workstations for assembly tasks should be
designed so that any operator can adjust to
his/her comfort to relieve stress and improve
performance. The ergonomically designed
smart assembly workstation is a solution to
ergonomic and productivity problems in the
workplace.
4. Female (women) are more productive than
male (men).
5. Creating a regression model representing
operator performance (productivity) was
built based on the experimental work.

The main contribution of this work is how to measure
the production rate of manual assembly lines based
on design ergonomically assembly workstation. The
author plans to conduct the future research in real life
case studies through validation this research in
different sectors of industries (manufacturing parts,
food industry and so on) and presented a new
performance measure for each specified operator in
these sectors.










REFERENCES
[1] Hasselquist, R.J., 1981. Increasing manufacturing
productivity using human factors principles.
Proceedings of the Human Factors Society 25
th

Annual Meeting, pp. 204-206.
[2] Schnauber, H., 1986. Ergonomics and
productivity as reflected in a new factory. Trends
in Ergonomics/Human Factors III, Karwowski
Ed., Elsevier Science Publishers, pp. 459-465.
[3] Ryan, J.P., 1989. A study of selected ergonomic
factors in occupational safety. Advances in
Industrial Ergonomics and Safety I, Anil Mital
Ed., Taylor and Francis, pp. 359-364.
[4] Das, B., 1987. An ergonomic approach to
designing a manufacturing work system. Int. J. of
Industrial Ergonomics. 1(3), 231-240.
[5] Resnik, M.L., Zanotti, A., 1997. Using
ergonomics to target productivity improvements.
Computers and Industrial Engineering. 33(1/2),
185-188.
[6] Burri, G.J., Helander, M.G., 1991. A field study
of productivity improvements in the
manufacturing of circuit boards. Int. J. of
Industrial Ergonomics. 7, 207-215.
[7] Shikdar, A., Das, B., 1995. A field study of
worker productivity improvements. Applied
Ergonomics. 26(1), 21-27.
[8] Das, B., Sengupta, A., 1996. Industrial
workstation design: A systematic ergonomic
approach. Applied Ergonomics, 27(2), 157-163.
[9] Das, B., Shikdar, A., 1999. Participative versus
assigned production standard setting in a
repetitive industrial task: a strategy for improving
worker productivity. Int. J. of Occupational Safety
and Ergonomics. 5(3), 417-430.
[10] Grandjean, E., 1988. Fitting the task to the man:
An ergonomic approach, Taylor and Francis,
London.
[11] Konz, S., 1995. Work design: Industrial
Ergonomics, 2nd edn, Grid Columbus, Ohio.
[12] Das, B., Grady, R.M. 1983. Industrial workplace
layout design: An application of engineering
anthropometry. Ergonomics. 26(5), 433-443.
Proceedings of the 41st International Conference on Computers & Industrial Engineering
281

[13] Salvendy, G., 1987. Handbook of Human
Factors. John Wiley, New York.
[14] Melamed, S., Luz, J., Najenson, T., Jucha, E.,
Green, M., 1989. Ergonomic stress levels,
personal characteristics, accident occurrence and
sickness absence among factory workers.
Ergonomics. 9, 1101-1110.
[15] Sanders, M.S., McCormic, E.J. 1992. Human
Factors in Engineering and Design, 6th edn,
McGraw Hill, New York.
[16] Wilson, J.R., Corlett, E.N., 1992. Evaluation of
Human Work: A Practical Ergonomics
Methodology. Taylor and Francis, Philadelphia.
[17] McLeod, D., 1995. The Ergonomics Edge:
Improving Safety, Quality and Productivity, John
Wiley, New York.
[18] Ayoub, M.A., 1990a. Ergonomic deficiencies: I.
Pain at work. J. of Occupational Medicine. 32(1),
52-57.
[19] Ayoub, M.A.1990b. Ergonomic deficiencies: II.
Probable causes. J. of Occupational Medicine.
32(2), 131-136.
[20] Shikdar, A., Al-Hadhrami, M., 2007. Smart
workstation design: an ergonomics and methods
engineering approach. Int. J. of Industrial and
Systems Engineering. 2(4), 363-374.































[21] Shikdar, A., Garbie, I., Khadem, M., 2011.
Development of a smart workstation for an
assembly task. Proceedings of the 2011
International Conference on Industrial
Engineering and Operations Management
(IEOM), Kula Lumpur, Malaysia, January 22-24,
2011.
[22] Yeow, P.H.P., 2003. Quality, productivity,
occupational health and safety and cost
effectiveness of ergonomic improvements in the
test workstations of an electronic factory. Int. J. of
Industrial Ergonomics 32(3), 147-163.
[23] Anon, A., 2005. Workstations, track system
smooth assembly. Assembly. 48(4), 32-35.
[24] Montgomery, D.C., 2009. Design and Analysis
of Experiments. 7
th
Edition, John Wiley and Sons,
Inc., Hobaken, NJ, USA.
[25] Montgomery, D.C., and Runger, G.C., 2011.
Applied Statistics and Probability for Engineers.
5
th
Edition, John Wiley and Sons, Inc., Hobaken,
NJ, USA.
[26] Navidi, W., 2008. Statistics for Engineers and
Scientists. 2
nd
Edition, Mc Graw Hill, New York,
NY, USA.

Proceedings of the 41st International Conference on Computers & Industrial Engineering
282