5.5 Energy efficiency in production processes – the influence of consumption
visualization and staff training
S. Asmus1, F. Karl2,3, M. Grassl3, A. Mohnen1, G. Reinhart2,3
1 Chair of Corporate Management, Technische Universitaet Muenchen, Germany
2 Institute for Machine Tools and Industrial Management, Technische Universitaet Muenchen, Germany
3 Fraunhofer IWU Project Group Resource Efficient Processing Machines, Augsburg, Germany
Abstract
This paper examines the influence of the visualization of consumed compressed air and staff training on the
consumption behavior of employees in a real production process. To measure potential changes in
consumption behavior a real-effort experiment at the Training Factory for Energy Productivity, a real
production setting at iwb of TUM, had been designed. Therefore, four groups were defined, each group in a
different experimental setting. This experiment is the first one ever conducted in a real-life setting and thus
adds valuable results to academia and practitioners. Compared to the group without any information about the
amount of consumed compressed air the participants provided with a display showing this information saved
on average 7-8%. The group provided with a movie about general measures to save compressed air in
production consumed around 24% less compressed air than all other groups of participants. Generally, no
significant differences between male and female participants had been found.
Keywords:
Empirical study, employee behavior, energy efficiency, production, sustainable manufacturing
1 INTRODUCTION
Today’s manufacturing companies are faced with the need to
reduce energy consumption sustainably [1]. Growing energy
prices [2] due to the increasing demand for energy are only
one reason. Moreover, in companies large energy saving
potentials that allow for increasing energy efficiency still exist
[3, 4].
In order to sensitize people for energy efficiency and show
possibilities to reduce energy consumption the Training
Factory for Energy Productivity (Lernfabrik für Energiepro-
duktivität, LEP) was built up at iwb (Institute for Machine
Tools and Industrial Management, see figure 1) [5].
Figure 1: Training Factory for Energy Productivity.
At LEP a small gearbox is manufactured. Therefore, the shaft
is turned, the main gear hardened by heating and quenching
and finally the gearbox assembled. To display the
manufacturing process machines of different ages, automatic
as well as manual processes and different forms of energy
(steam, electricity, thermal energy and compressed air) are
used.
During a sensitization training at LEP participants from all
hierarchical levels learn and practically apply a
methodological approach that can directly be utilized in real
production environments, the Energy Value Stream (EVS) [6].
EVS mainly consists of two phases: the analysis and the
design phase (see figure 2).
•Measurement
•Visualization
•Analysis
System Elements Types of Energy Waste
Energy Value Stream Analysis
•Generation of optimization measures
•Prioritization
•Identification of interactions
Freedom of Action Design Toolbox
Energy Value Stream Design
Implementation
Figure 2: Energy Value Stream (EVS).
G. Seliger (Ed.), Proceedings of the 11th Global Conference on Sustainable Manufacturing - Innovative Solutions
ISBN 978-3-7983-2609-5 © Universitätsverlag der TU Berlin 2013
181
S. Asmus, F. Karl, M. Grassl, A. Mohnen, G. Reinhart
EVS was deduced from the methodology value stream
mapping from Lean Management [7]. During the analysis
phase energy waste is identified by measuring energy
consumption in a defined area, visualizing the values and
applying various analysis methods. Thereby, the three
system elements (technology & system, organization &
management, human & behavior [8]) and different types of
energy waste (overproduction, dead time, transport,
inventory, rejections, movements, unused potential of
employees) have to be considered. The design phase aims at
limiting energy waste. For this purpose the freedoms of action
have to be defined and optimization measures generated by
applying a design toolbox. Then measures are prioritized
regarding their complexity and cost effectiveness. After
choosing the right measures, they have to be implemented.
When optimizing production technological systems the three
already mentioned system elements need to be considered.
Since numerous works in the fields of the first two elements
were already carried out [9] this article focuses on human &
behavior. Furthermore, workers in production have due to
their behavior a large influence on energy consumption.
Therefore, a study was carried out to analyze their influence.
For this purpose a process step at LEP was chosen where
workers’ behavior affects energy consumption. Hence, the
final assembly station was picked. Here, the gear box is
screwed pneumatically by 6 bolts. Another reason for
choosing this process was the use of compressed air, as the
economical application of compressed air is crucial due to its
poor degree of efficiency. The worker can influence the
consumption of compressed air by setting the pressure at the
workplace.
The findings of the study will be presented in the following
chapters. It was conducted in an interdisciplinary team
consisting of engineers and behavioral economists.
2 STATE OF THE ART
Even though the public discussion about resource efficiency,
environmental issues and climate protection increased
tremendously over the past years [10] only a few studies on
energy efficient measures in the work-place context have
been conducted so far [11]. One of the few studies in that
field is the work of Siero et. al. about the influence of goal-
setting, feedback and education on employees’ behaviour
[12]. They figured out that among other things creating
awareness for the topic of energy efficiency as well as goal-
directed education and feedback lead to significant behaviour
changes of the workforce resulting in less energy-wasting.
In order to enhance the available findings on how to increase
resource efficiency in the work place established concepts
from the field of behavioural economics should be applied
[13, 14]. Therefore, this study will put strong emphasize on
the feedback mechanism consumption visualization and staff
training also as potential measures to increase energy
efficiency in production processes.
3 SETUP OF THE STUDY
The experiment took place at LEP in November and
December 2012. In total 160 students took part in the study
and were randomly distributed to the four different conditions
of the experiment. The experiment took between 45 and 60
minutes for each participant and they were remunerated with
a fixed payment of 8 euros. In each experimental group
consisting of 40 students 13 had been female and the other
27 male. Therefore, an equal gender distribution over the
groups is guaranteed. The general experimental setting can
be seen in figure 3.
Figure 3: Experimental setting.
Before the experiment started participants were introduced to
the work station by a power point presentation and a video to
ensure a standardized procedure for every participant.
Following, all participants got a five minutes lasting trial round
to get familiar with the work place setting and the task. After a
short break with additional information people started with the
first round which took 10 minutes. Depending on the group
the students belonged to a certain movie was shown to them
which had duration of around five minutes. Group C and T1
saw a movie about a new faculty at Technische Universitaet
Muenchen (TUM), the TUM School of Education. The movie
had no relation to the task, the environment or energy saving
information. Group T2 got a movie showing nature scenes to
address the environmental awareness of the participants. To
group T3 a movie was shown which gave particular
information on how to reduce consumption of compressed air
in production. After the movie participants did execute the
second round of the experiment for ten minutes. The last step
of the procedure was a questionnaire which had to be filled in
by all participants.
As it can be seen in figure 3 groups T1, T2 and T3 had an air
flow meter next to them on the work station during the whole
duration of the experiment. Therefore, they were able to get
continuous information about their cumulated consumption of
compressed air.
After each of the three rounds the experimenter counted the
finished and unfinished gear boxes the participant performed.
This number was after the experiment compared to the used
amount of compressed air by each round to calculate the
exact number of litres of compressed air per screwed bolt
(l/bolt).
4 RESULTS
4.1 Differences between experiment rounds and
treatment
First of all the influence of visualizing consumption of
compressed air on participants’ behavior is shown. To isolate
the effect of the display on the consumption only the results
of the trial round and the first round are taken into
182
Energy efficiency in production processes – the influence of consumption visualization and staff training
consideration since besides the display’s appearance for
groups T1 – T3 and non-appearance for the control group
everything is equal over all four groups in these two rounds.
As it can be seen in table 1 where the results are ordered by
the experiment sequence in the trial round group C uses on
average 10.76 l/bolt and the treatment groups between 9.87
and 10.18 l/bolt. This results in a saving between 5.4% and
8.3% per group and 7.2% in average over all three groups in
the trial round only due to the display. Having a closer look on
the first round the savings related to the visualization of the
energy consumption are between 6.6% and 9.7% per group
and on average 7.6% over the three groups with a display
compared to the control group.
Round
Group
N
Mean
(l/bolt)
SD
(l/bolt)
T
C
40
10.76
0.85
T1
40
10.18
1.15
T2
40
9.87
1.04
T3
40
9.90
0.92
1
C
40
10.68
0.67
T1
40
9.98
1.19
T2
40
9.64
1.25
T3
40
9.98
1.09
2
C
40
10.72
0.72
T1
40
10.08
1.39
T2
40
9.60
1.38
T3
40
7.54
1.41
Table 1: Energy consumption within the three rounds.
Table 2 illustrates the mean consumption of compressed air
ordered by group. Interestingly no noteworthy learning effects
in terms of energy efficiency can be seen when comparing
the mean consumption per bolt between the periods for every
single group. This is an important finding because occurring
differences between the groups and periods will be based on
the different treatments and not on potential learning effects
regarding the usage of compressed air.
While the consumption of the groups C, T1 and T2 remains
relatively constant over time the consumption of group T3
drops from round one to round two by 24.4%. This implies
two major findings. First of all, the purely confrontation of the
participants with a video showing nature sceneries to build
environmental awareness as done with group T2 has no
impact on the energy consumption behavior of people. Only
staff training on how to save energy while doing a certain
task, not related to any environmental issue, leads to
significantly decreasing energy consumption as it can be
seen in the results of group T3. As expected the movie which
was unrelated to the whole experiment and presented to the
control group and T1 had no influence on participants’
behavior.
Group
Round
N
Mean
(l/bolt)
SD
(l/bolt)
C
T
40
10.76
0.85
1
40
10.68
0.67
2
40
10.72
0.72
T1
T
40
10.18
1.15
1
40
9.98
1.19
2
40
10.08
1.39
T2
T
40
9.87
1.04
1
40
9.64
1.25
2
40
9.60
1.38
T3
T
40
9.90
0.92
1
40
9.98
1.09
2
40
7.54
1.41
Total
T
160
10.18
1.05
1
160
10.07
1.13
2
160
9.48
1.73
Table 2: Energy consumption within the four groups.
To get a deeper understanding of the discussed findings the
boxplot in figure 4 visualizes the results, differentiating
between the three rounds of the experiment and additionally
between the four groups. What becomes very obvious here is
that the energy consumption of different people varies
considerably. While the 25th percentile (lower quartile), the
75th percentile (upper quartile) and especially the medians
are rather similar over time for groups C, T1 and T2 the
consumption of group T3 in the second round is strongly
affected by the additional information on energy saving and
therefore drastically lower as discussed above.
610 12 14
48
liters/bolt (l/bolt)
CT1 T2 T3
Group
Trial Round (T) First Round (1)
Second Round (2)
Figure 4: Boxplot of the consumption distribution.
In order to figure out if the consumption differences between
the four conditions in that experiment are statistically
significant the results of a Bonferroni test for each of the
three rounds was executed. Based on the number of
experimental groups a multiple comparison of the means
183
S. Asmus, F. Karl, M. Grassl, A. Mohnen, G. Reinhart
between all groups is done in table 3. In this table the means
of the consumption are compared group by group and the
differences are shown with a positive or negative sign in front
of the mean value difference. In the trial round group T1
which had the visualization on the consumption uses on
average -0.575 l/bolt less than group C who had no feedback
on the energy usage. Additionally to the mean savings per
round measured in l/bolt the related significance levels are
shown in the table always below the number of the mean
savings. In this example p=0.065 and with p>0.05 not
significant on the 5% level. Therefore, the difference in this
comparison is not statistically significant.
Trial Round
Row Mean –
Column Mean
C
T1
T2
T1
deviation
p-value
-0.575
0.065
T2
deviation
p-value
-0.8875
0.001
-0.3125
0.978
T3
deviation
p-value
-0.8525
0.001
-0.2775
1.000
0.035
1.000
First Round
Row Mean –
Column Mean
C
T1
T2
T1
deviation
p-value
-0.695
0.026
T2
deviation
p-value
-1.04
0.000
-0.345
0.917
T3
deviation
p-value
-0.6975
0.025
-0.0025
1.000
0.3425
0.935
Second Round
Row Mean –
Column Mean
C
T1
T2
T1
deviation
p-value
-0.635
0.155
T2
deviation
p-value
-1.1175
0.001
-0.4825
0.536
T3
deviation
p-value
-3.18
0.000
-2.545
0.000
-2.0625
0.000
Table 3: Comparison of the mean consumption (in [l/bolt]).
By taking a closer look on the results of the second round it
can be seen that the mean consumption of group T3 is 3.18
l/bolt lower compared to the control group. Below that value
the p-value is given. The related p-value to the value 3.18
l/bolt is 0.000, and with p < 0.001 highly significant. For the
second round of the experiment all p-values of T3 compared
to the other groups are 0.000 and therefore highly significant
on the 1% level. This supports the findings of the descriptive
comparison of the means for the second round in table 1 and
2 as seen above.
4.2 Gender differences
Because of the fact that the number of female participants in
the experiment is equally distributed over the four groups the
consumption between male and female students can easily
be compared. The results in figure 5 show that the
consumption levels of both genders are nearly at the same
level comparing every single round and every single condition
separately. Female participants on average over all groups
consumed 9.43 l/bolt and therefore a little less than their
male counterparts who consumed on average over all groups
9.51 l/bolt.
024610
liters/bolt (l/bolt)
CT1 T2 T3
Group
Female Male Female Male Female Male Female Male
Trial Round (T) First Round (1)
Second Round (2)
Figure 5: Consumption comparison by gender.
4.3 Goal changing behavior
In the questionnaire after the experiment students were
asked to name their major goal for each of the two rounds.
They had to choose between either a) produce high
quantities (Quan), b) avoid mistakes (Qual) or c) save energy
(Energy). Table 4 shows separately for round one and round
two the answers of the participants, differentiating between
the four experimental groups. The numbers in brackets show
the percentages of students per group which chose a
particular goal.
For groups C, T1 and T2 it can be seen that the number of
students who named as their major goal to produce high
quantities rose from round one to round two tremendously.
Over the three groups the percentage increased from 33.9%
to 73.5%. In comparison the number of students who were
trying to avoid mistakes or save energy decreased strongly in
these groups between the rounds.
184
Energy efficiency in production processes – the influence of consumption visualization and staff training
Round
Group
N
Quan
Qual
Energy
1
C
40
16
40%
24
60%
0
0%
T1
40
13
33%
17
42%
10
25%
T2
38
11
29%
20
53%
7
18%
T3
39
23
59%
16
41%
0
0%
Total
157
63
40%
77
49%
17
11%
2
C
40
34
85%
6
15%
0
0%
T1
40
27
68%
8
20%
5
12%
T2
37
25
68%
6
16%
6
16%
T3
39
15
38%
3
8%
21
54%
Total
156
101
65%
23
15%
32
20%
Table 4: Change of participants’ main goal between rounds.
In contrast, participants of group T3 changed their behavior in
a different direction. More than 50% of them were looking
mainly on reducing energy consumption in round two, while
none of them called energy savings the main goal in the first
round.
Based on these results it becomes obvious that in case
people get more confident and familiar with a certain task
they tend to focus more on producing high numbers while
taking less the quality and the energy consumption into
account. In contrast to that people who get a certain external
impulse on how to change behavior related to energy
efficiency, these people do focus more on that goal
dimension. These findings are supported by the comparison
of the increase of inserted bolts. While all groups completed
on average 72 – 74 bolts in the first round the groups C, T1
and T2 realized 79 – 82 bolts in the second round while T3
grew only slightly from 74 to 75 bolts in the second round.
4.4 Results summary
To sum up the most important findings of the study are:
Energy can be saved only by visualizing the
consumption.
General sensitization regarding environmental awareness
has no effect on behavior.
Workers have to be sensitized and trained on the specific
topic to behave in a more energy efficient way.
Between females and males no significant behavior
differences related to energy saving behavior exist.
Even without financial incentives people do change
behavior based on additional information.
5 CONCLUSION AND OUTLOOK
This paper presented a study to analyze the influence of
energy consumption visualization and task-related
information on workers’ behavior. To conduct this experiment
a work station to assemble gear boxes with a pneumatic
screw driver was chosen and the behavior of 160 participants
analyzed. Four different groups consisting of 40 participants
each were defined and separated in different treatments.
Generally, the strength of the influence of workers’ behavior
on the energy usage in a certain production step became
obvious. The most important findings are that simply showing
the consumption of compressed air during the production
process to the worker reduces the consumption by around
7%. By giving additional task-related training with a focus on
saving energy participants reduced the compressed air
consumption by additional 24%. There haven’t been found
any significant differences between the results of female and
male participants.
Future research should first replicate the scenario in a
completely real production setting in industry to validate the
results of that experiment. Furthermore, other related topics
should be tested in the LEP-setting to gain further insights on
human behavior and the reaction on consumption
visualization, additional task-related information or other
related topics to enable and foster energy efficient behavior.
To ensure that the experimental setting is as close as
possible to a real production environment the three main goal
dimensions in production settings namely energy efficiency
(in broader terms material efficiency), product quality, and
produced quantity have to be taken into account jointly.
6 ACKNOWLEDGMENTS
We extend our sincere thanks to the Free State of Bavaria for
funding the project Green Factory Bavaria in the framework of
the future initiative "Aufbruch Bayern".
7 REFERENCES
[1] Abele, E., Reinhart, G., 2011, Zukunft der Produktion,
Hanser, München.
[2] Kleiner M., Tekkaya A. E., Becker D., Pietzka, D.,
Schikorra, M., 2009, Combination of curved profile
extrusion and composite extrusion for increased light-
weight properties, Production Engineering - Research
and Development, 3:63-68.
[3] McKinsey (ed.), 2009, Energy: A key to competitive
advantage, Frankfurt.
[4] Neugebauer, R., Westkämper, E., Klocke, F., Kuhn, A.,
Schenk, M., Michaelis, A., Spath, D., Weidner, E.,
2008, Untersuchung zur Energieeffizienz in der
Produktion, Fraunhofer Gesellschaft, München.
[5] Reinhart, G., Karl, F., 2011, Live Experience of Energy
Productivity – The Training Factory at Technische
Universitaet Muenchen (TUM), 1st Conference on
Learning Factories, Darmstadt.
[6] Reinhart, G., Karl, F., Krebs, P., Maier, T., Niehues, K.,
Niehues, M., Reinhardt, S., 2011, Energiewert-
stromdesign, wt Werktstattechnik online, 4/101:253-
260.
[7] Rother, M., Shook, J., Womack, J.P., 2000, Training to
See, Lean Enterprise Institute, Brookline.
[8] Bullinger, H.-J., Spath, D., Warnecke, H.-J.,
Westkämper, E., 2009, Handbuch Unternehmens-
organisation: Strategien, Planung, Umsetzung,
Springer, Berlin.
185
S. Asmus, F. Karl, M. Grassl, A. Mohnen, G. Reinhart
[9] Duflou, J.R., Sutherland, J.W., Dornfeld, D., Herrmann,
C., Jeswiet, J., Kara, S., Hauschild, M., Kellens, K.,
2012, Towards energy and resource efficient
manufacturing: A processes and systems approach,
CIRP Annals – Manufacturing Technology, 61:587-609.
[10] Belz, F.-M., Peattie, K., 2009, Sustainability Marketing:
A Global Perspective, John Wiley & Sons, Chichester.
[11] Carrico, A.R., Riemer, M., 2011, Motivating Energy
Conservation in the workplace: An Evaluation of the
Use of Group-Level Feedback and Peer Education,
Journal of Environmental Psychology, 1/31:1-13.
[12] Siero, F.W., Bakker, A.B., Dekker, G.B., van den Burg,
M.T.C., 1996, Changing Organizational Energy
Consumption Behavior Through Comparative
Feedback, Journal of Environmental Psychology,
3/16:235-246.
[13] Tversky, A., Kahneman, D., 1973, Judgement under
Uncertainty: Heuristics and Biases, Science,
September /185:1124-1131.
[14] Tversky, A., Kahneman, D., 1981, The Framing of
Decisions and the Psychology of Choice, Science,
January /211:453-458.
186
