The Good, the Bad and the Mediocre: C reating Insightful Stories on
Process Improvement
Markus Salge
Industrieseminar, Mannheim University
Schloss, D-68131 Mannheim, Germany
phone: +49 621 181 1585
facsimile: +49 621 181 1579
email: salgem@is.bwl.uni-mannheim.de
web: http://is.bwl.uni-mannheim.de/
Keywords: process improvement programs, generic models, dynamic stories on process im-
provement initiatives
Abstract: Building upon previous work in the field of system dynamics, a generic model of multi-
ple improvement initiatives is outlined. The model is used to create insightful stories on success
and failure in process improvement initiatives. The simulation experiments reveal that plants
should strive for implementation patterns that focus on programs exhibiting higher organiza-
tional complexity rather than technical complexity. Furthermore, the simulation analyses provide
insights in the interplay between organizational learning, program commitment, and process
improvement. The value of the conducted approach lies in the explicit investigation of the impact
of varying improvement program patterns on plant performance.
Introduction
According to Jay W. Forrester, system dynamics “is a way of studying the behavior of [dynamic]
systems to show how policies, decisions, structure, and delays are interrelated to influence
growth and stability” (Forrester 1961: vii). Working with graduate students in the field of system
dynamics, the author feels that quite too often students spend a lot of effort on explaining the
model structure but comparatively little on the discussion how the structure of a model influences
its behavior. One might get the idea that students seem to avoid this task even if they are asked to
equally spend effort on system thinking, modeling, and simulation. The reason to this might be
that students are more confident in explaining model structure due to its formal form then in ana-
lyzing why a certain pattern of behavior occurs. This might also be the case as methods of formal
mathematical behavior analysis, e.g. on loop dominance (Richardson 1995), are still the domain
of a few experts in that field (e.g. Ford 1999; Mojtahedzadeh, Andersen, and Richardson 2004;
Kampmann and Oliva 2006; Gineralp 2006). In addition, user-friendly software tools like “Di-
gest” (Mojtahedzadeh, Andersen, and Richardson 2004) are still experimental and thus only ap-
plicable to a limited extent. Thus, a one-click application for behavior analysis is not available up
to date. However, in the author's opinion one might read a lot of interesting stories from the be-
havior of a system even with the naked eye, even though it often takes an experienced system
dynamicist to reveal the dominating loops from model behavior (cf. Kampmann and Oliva 2006).
One challenge to education in system dynamics may lie in teaching students to read those insight-
ful dynamic stories which models tell us.
To foster this idea of dynamic story telling, this article focuses on the behavior of a generic
process improvement model, which has been generated in order to provide insights to success and
failure of multiple improvement initiatives in manufacturing. This is part of an ongoing research
project. The model will be introduced in the next section. Subsequent to this, three different
simulation experiments are outlined in broader detail. The article ends with a discussion of the
results and with an outlook on subsequent research.
Quantitative and qualitative modeling approaches
In spite of its early entry into system dynamics, the concept of generic structures is still develop-
ing. Based on Forrester’s notion of “general purpose models” (Forrester 1961: 313), the concept
of generic structures has evolved mainly into the branches of quantitative and qualitative models
(Coyle 2000; Liehr 2004, 2001). The former type includes “generic (canonical) situation models”
and “abstracted micro-structures”, the latter “counterintuitive system archetypes” (Lane and
Smart 1995). Forrester’s “Market Growth as Influenced by Capital Investment” (1968) or Lyneis’
“Corporate Planning and Policy Design” (1988) are examples of generic models. They are the
formal representation of a problem and structure common to many situations. These models—
contrary to micro structures— are not designed as building blocks for larger models. Micro struc-
tures differ from generic models in both the extent of their structure and their transferability into
other contexts. Due to their high aggregation, they can be applied to other situations as building
blocks. Micro structures can be classified into those which serve as building blocks to structures
from certain areas and into those which are applicable in many different contexts (Paich 1985).
As building blocks of systems, micro structures can facilitate understanding of complex interac-
tions in social systems (Milling 1972). The second branch of generic structures— system arche-
types— are mainly based on Meadows’s (1982: 98) “persistent, system-dependent malfunctions”
and on Senge’s (1994) monograph “The Fifth Discipline”. Senge especially emphasizes the ge-
neric characteristics of his nine archetypes which can provide an explanation to counterintuitive
behavior in different contexts. The value of system archetypes lies especially in their limited ex-
tent and their transferability to recurring system behaviors.
The same categorization in qualitative and quantitative approaches can also be found in pre-
vious work in the field of system dynamics analyses on process improvements. As an example of
the former, Carrol, Sterman, and Marcus (1997) use a case study at Du Pont for their investiga-
tion on proactive maintenance programs. They use a qualitative system thinking approach with-
out explicit system dynamics modeling, although they use level-rate-diagrams for model illustra-
tion (cf. Sterman 2000). They outline a typical fixes-that-fail-archetypical behavior, i.e., that less
proactive maintenance activities increase productivity in the short run but decrease in the long
run, due to the increasing equipment downtime. Repenning and Sterman (2001), Keating et al.
(1999), Repenning and Sterman (1997) as well as Oliva, Rockart, and Sterman (1993) abstract
from specific improvement programs and analyze process improvement programs more generally
with system thinking as methodology. All four articles base on case studies from multiple im-
provement programs examined at different sites. Beside other valuable findings, they outline that
improvement initiatives can facilitate subsequent improvement efforts, if they are evaluated as
successful by both managers and workers. However, in the case of low perceived success the
same interrelation can also hinder continuous process improvements. Kim (1993) provides two
case studies upon process improvement programs (total quality management [TQM; cf. Shiba,
Graham, and Walden 1993] and product development management) in which Senge’s system
archetypes have been applied in order to facilitate organizational learning.
As an example of quantitative approaches, Sterman, Kofman, and Repenning (1997) analyze
a TQM program at Analog Devices and provide a fully documented system dynamics model
(documentation is available in Repenning and Sterman 1994). In their case study with Analog
Devices they reveal that due to Analog’s TQM program the productivity grew faster than cus-
tomer demand and did thus generate excess labor capacity and massive layoffs. The authors pro-
vide an extensive model which is highly specific to the Analog case. In spite of the great value of
their work to management literature, the transferability of the model is therefore limited. Other
formal modeling approaches on process improvement programs have been conducted by Repen-
ning (2002, on TQM) and Maier (2004; 2000, both on total productive maintenance [TPM;
cf. Nakajima 1988]). Even though both authors provide mathematical equations to some model
interrelations, they do not include a complete model listing. In contrast, Thun (2006) analyzes the
interplay of different components of TPM and provides all model equations in his article. For this
purpose he expands Sterman’s (2000) proactive maintenance model by further components that
are specific to the TPM approach (e.g. autonomous maintenance and maintenance prevention).
His insightful analyses are very specific to the TPM approach and therefore it is only possible to
generalize his results to a limited extent to the implementation of other process improvement
programs, which is aimed by the study outlined in this article.
A generic, quantitative model of multiple process improvement pro-
gramms
Building upon both qualitative and quantitative approaches, a generic model of multiple im-
provement initiatives is outlined. Existing micro structures are applied as building blocks where
possible (e.g. from Hines 2005, Sterman 2000, Repenning and Sterman 1994, and Lyneis 1988).
The model is intended to provide insights in several program implementation patterns. This is
necessary as different pattems show varying success in plant performance (Filippini, Vinelli, and
Voss 2001). Figure 1 gives a brief overview of the model structure:
Management hiring / Laying off
+ commitment
+ support & training management focus training
Improvement ieartitg trom Human Resources
Programs Improvements “labor:
+ workers’ improvement
experiences
+ quality management & orkers’ improvement
control ee Tey « workers’ trainings level [7
+ maintenance hiring & layoffs
+ process acceleration workers’ productivity) . workers’ program
+ labor productivity improvement commitment
results mi
quality] — workers’lmachinery]process| processing suppliers’ labor
controll productivity} up-time| quality| time] quality eos
Z demand =
Manufacturing Market & Finance
System |_ capital costs
y materials costs _,| * Perceived quality
+ perceived lead time
+ backlog defective parts _,| + perceived price ratio
+ material inventories + revenues / costs
+ work in process ‘on-time deliveries ,| -
{Marin process: financial accounting
materials & products
+ machinery capacity
labor capacity
process yield financial stress / market yield
Figure 1: Overview of model structure
The model consists of five sectors. The model equations can be found in the supplementary
file. In the human resource section of the model, hiring and laying-off of workers is conducted
according to the perceived labor productivity and desired gross production rate. The latter is de-
rived from customer demand, which means that low (high) workers’ productivity and compara-
tively high (low) demand leads to hiring (laying-off) of workers (Hopp and Spearman 2001). The
training level of the workers depends on on-the-job training provided by management
(Armstrong 2003). Contrary to that, workers’ improvement experiences cannot be controlled di-
rectly by management in the model. Management can only provide free time to the workers to
gain experiences with process improvements, as it is the case in the concept of Kaizen (Imai
1986). But how the workers use this freedom and benefit from it mainly depends on their pro-
gram commitment (Armstrong 2003). The program commitment deteriorates if the workers per-
ceive a low job security (Meyer and Herscovitch 2001) and it increases if the improvement initia-
tives show to be successful (Meyer and Allen 1991). Furthermore, the workers’ program com-
mitment depends on the perceived management support which is necessary for the improvement
initiatives (Senge 1999). An empirical investigation conducted by Neubert and Cady (2001)
shows that the factors job security, program success, and management support have significant
impact on workers’ program commitment, and that in tum workers’ program commitment is
leading towards higher workers’ effort for improvement programs. This is also underpinned by
the findings of Sterman and Repenning (2001) et a/., examining improvement initiatives at dif-
ferent plants. Management (see management sector) will only provide support to the workers if
they also evaluate the improvement initiatives as successful. In the model, management's pro-
gram commitment therefore depends on both, perceived improvement results and improvement
expenses and financial stress, respectively (Repenning and Sterman 2001, et a/.).
The market and finance sector exhibits three figures for plant performance: the ‘perceived
lead-time’ for time, ‘perceived price ratio’ for costs, and ‘perceived quality’ for quality. The plant
loses and gains market shares pursuant to its performance in comparison to its competitors (Hill
2000). Costs per units are determined by the plant’s material, labor and capital costs (Milling
1974). The price is calculated with a profit margin over unit costs, which is endogenous and
changes according to a desired market share (Hanson 1992). The three performance figures for
cost, quality, and time change according to the price for products, the fraction of delivered defec-
tive parts, and the order lead-time. The latter two are determined in the manufacturing system.
perceived
management + improvement results job security
foci \ fr +\ (+ management
support
workers’
)
pas
machinery commitment
downtimes
/
fa workers’
& workers’ | improvement
improvement experiences | ‘imnd tor
effort
improvements
; a ; training
available workers’ training
machinery level
Figure 2: Interactions between management foci, workers’ commitment and
defects levels
NXT
C
In the manufacturing system, materials are processed through the production stations and in-
ventories (also for the following, cf. Hopp and Spearman 2001). Defective materials might be
delivered from suppliers or can occur due to inadequate production processes. Some of the defec-
tive parts are detected through Quality Control (cf. Ishikawa 1985) but some are delivered to the
customer, which deteriorates the quality reputation of the plant. In addition, the production capac-
ity depends on both machinery uptime and labor capacity. Besides available production capacity,
the production lead-time of the manufacturing system is also depending on the machinery proc-
essing time.
Every constraint in the production system— suppliers’ quality, process quality, quality con-
trol, machinery uptime, labor productivity, and processing time— is represented by a level of de-
fects in the model (see improvement sector). In this article, the term ‘defect level’ is used in its
most general sense according to Schneiderman (1988: 53), like “errors, rework, yield loss, [... ]
unscheduled downtime, [... ] poor quality”, and so on. In the model, each defect level is the target
of an improvement initiative, as illustrated in Figure 2, showing the main connections between
the sectors ‘management’, ‘human resources’, ‘improvement programs’, and ‘manufacturing sys-
tem’. Schneiderman (1988) found in an empirical investigation that experienced improvement
teams maintain a constant improvement rate, i.e., the level of defects exhibits a similar behavior
as radioactive decay, which means that the amount of time necessary for a level of defects to drop
by 50% is constant. In addition, Schneiderman revealed that the constant half-life time (¢) in-
creases according to organizational and technical complexity of the improvement effort. Schnei-
derman found that initiatives which are placed in the left bottom part of the matrix in Figure 3
exhibit half-life times of approximately one month and in the right upper part of twenty-two
months. An improvement initiative in suppliers’ quality— for example— involves people from
different functions and organizations and thus possesses high organizational complexity. Con-
trary to that, the dimension of technical complexity grasps the novelty of the applied technology
and therefore reductions in processing time feature a higher technical complexity than improve-
ments in suppliers’ quality. The adopted Schneiderman-Matrix with an indication of each im-
provement initiative incorporated in the model is illustrated in Figure 3.
increasing technical complexity
£ Multiple improving suppliers’ %
3 Organizations Suallty 3
. a
EF :
5 improving quality 8
[s) Of processes downtime S
Bw (Cress reductions 2
£ ; increasing labor 3
& Functional 9 2
2 productivity | &
| a processing time |] ©
a improving quality reductions o
E+ control
) Single 8
6 Function 8
low high
Technical Complexity
Figure 3: The Half-Life/Complexity Matrix adapted from
Schneiderman 1999, 1988)
Due to increasing half-life times, a plant has to allocate more efforts to complex than to sim-
ple programs in order to achieve improvements. For example, improvements in labor productivity
can be achieved with little effort, comparatively. However, increases in labor productivity do not
necessarily stimulate demand. Therefore, high improvement rates in labor productivity can lead
to excess capacity if demand is not increasing at the same rate. Thus, plants should also engage in
improvement efforts that upgrade the plant’s performance in ‘order qualifying’ and ‘order win-
ning’ criteria, respectively, like quality and time (Hill 2000). Lower costs due to higher produc-
tivity might not be sufficient to generate higher demand, if price is just an ‘order qualifying’ cri-
terion. Schneiderman (1999) also emphasizes that the half-life times outlined in his matrix can
only be achieved by an experienced improvement team and that not every plant will be able to
achieve such improvement rates right from the start. He suggests that plants with low experience
in process improvements should start with less complex initiatives which can contribute to organ-
izational learning (cf. Stata 1989). Gains in process improvement experience facilitate the plant’ s
capabilities to handle higher organizational and technical complexity, and from that the plant can
strive for more ambitious improvement efforts. This process of organizational learning with the
interplay between workers’ improvement experiences, gains in machinery up-time, and process
yield is illustrated in Figure 2.
Transforming Schneiderman’s original equation to an integral form (1988, 1999), a level of
defects (¥;), its improvement (imp;) and deterioration rate (der) can be calculated according to’
. . In(2) In(2)
Y=%, +f (det; - imp;) dt; imp, = (Yon, 1* OAs det;= * (Yona, —¥) +
bon te,
: i focus on
(Yirax) Maximum (Yoip) minimum (ajmanagement workers’
machinery doses mathinery downtime machinery down-time _pogen
8 _ “commitment
oO Ve > (Y) Machinery
(det) deterioration Down-time r
in machinery up-time 7 imp) improvements ing——______—_(R)workers'
} machinery down-time improvement effort to.
' (Yo) initial machinery programs
(te) erosion fime down-time (ta) halflife time
machinery up-time
average workers’ skill
and training
Figure 4: Representation of machinery down-time in the improvement pro-
gram sector
One shortcoming in Schneiderman’s concept is his explicit assumption that the waste reduc-
tion rate (i.e., the improvement rate) is given and constant over time (cf. Dutton and Thomas
1984). This implicitly implies the underlying leaming rate to be exogenous and thus, independent
of managerial efforts, workers’ commitment and gains in experience. Contrary to that, Lapré er
al. (2000) show in a longitudinal empirical investigation that improvement rates are changing
over time in accordance to managerial efforts and a continuous process of learning. In order to
incorporate these findings, Schneiderman’s original improvement rate equation is supplemented
with the factors «; for management focus for defect level i and # for program commitment, work-
ers’ skills and training level. Therefore, if management is solely focusing on improvements in
defects level i (a;=1) and workers are highly experienced and motivated (f=1) the plant will yield
the same improvement rate as outlined in the half-life/complexity-matrix. On the other hand, if
management and workers do not make a sufficiently high effort towards maintaining process im-
provement, the defect level deteriorates to its maximum value.
Figure 4 illustrates the stock and flow structure of the defect level ‘machinery down-time’,
which stands for an equipment maintenance initiative. Improvements are represented with an
outflow and deteriorations with an inflow, respectively. The other improvement initiatives are
modeled correspondingly, with specific initial values, half-life times, erosion times, and man-
agement foci towards improvement. In its initial state, the model is set into equilibrium", which
means that the management foci on the varying improvement initiatives are adequate in order to
maintain the defects levels and the market share goal in accordance to the workers’ program
commitment, training and experience. Without any adjustments to the different foci, the market
share goal, or the customers’ expectations on quality, time, and costs, the plant maintains its
status quo. This case is referred to as ‘The Mediocre’ in this article. In the following, two differ-
ent insightful dynamic stories are being discussed.
Insightful stories on process improvements
Inspired by the plot of Sergio Leone’s (1966) epic ‘spaghetti’ western “The Good, the Bad and
the Ugly” of three men seeking a fortune in buried coins, the three different dynamic stories on
process improvement are referred as ‘the Good’, ’the Bad’ and ’the Mediocre’, whereas the latter
serves as a base run. In the plot outlined in this article, the Good and the Bad strive for market
shares, which means that they increase their desired market shares by a constant rate over approx.
3 years. In order to stimulate customer demand, both protagonists undertake improvement initia-
tives, but choose different pattems of process improvement. The Good takes a path with com-
paratively high organizational and the Bad with high technical complexity, respectively, as can
been seen in Figure 5:
12 Organizational complexity
The Good
The Mediocre +
The Bad -
1 Technical complexity
(ee The Good
The Mediocre
TheBad ==
Month
©
0 Year 1 Year 2 Year 3 Year 4 Year 5
Time (Days)
Figure 5: Organizational & technical complexity
Therefore, the Good is shifting his focus rather to the left upper part and the Bad to the right
lower part in the half-life/complexity matrix in Figure 3. Nevertheless, even if the Good takes a
‘softer’ and the Bad a ‘harder’ improvements approach, they do not neglect other initiatives com-
pletely. This plot outline has been chosen since softer approaches showed to be more successful
in means of process improvement than harder approaches (Filippini, Vinelli, and Voss 2001;
Vargas and Cardenas 1999). These empirical findings serve for testing the behavioral validity of
the model. It should be noted that according to Schneiderman organizational improvement pro-
grams imply comparatively higher complexity than technical initiatives. Thus, in Figure 5, the
organizational exceeds the technical complexity in the equilibrium run (complexity is measured
in months for half-life time).
In the dynamic stories outlined in the following, the Good and the Bad are changing their fo-
cus once and maintain it until the end of the simulation runs. It should be noted that both pro-
tagonists spend the same overall effort on process improvement initiatives, but with different
foci. Except for their increasing desired market shares, all other parameters stay at their initial
values.
Market Share Labor force
Worker
0 Year 1 Year2 —_Year3 Year 4 Year 5 ) Year 1 Year 2 Year 3 Year 4 Year 5
———— “Time (Days)
Figure 6: Market share and labor force
As can been seen in Figure 6, the Good is gaining market shares, while the Bad is constantly
losing. This is due to the customers’ different perception of their performances in quality, cost,
and time. For example, the Good's performance in quality (see Figure 7) is increasing due to his
higher focus on suppliers’ quality, while the Bad undertakes more initiatives in process accelera-
tion and machinery up-time. Gains in quality exhibit a beneficial side effect in the model (and
reality), as they increase the net production rate while reducing scrap and inventories. Thus, the
Good’s unit costs are decreasing and the Bad’s are increasing, respectively (see Figure 8). This
increase in net production rate is also causing the Good’s labor force to decline for approx. 3
months, until the effect from the rising market share develops its momentum and the labor force
starts to increase.
Perceived quality relative to competitor Perceived price relative to competitor
Dmni
Dmni
0.95 og
0 Yeart Year2 Year3 Yard. Year 0 Yeart Year2“Year3. «= Year4. = Year
nae The Good ©
The Mediocre
TheBad —
Time (Days) Time (Days)
Figure 7: Perceived performance in quality and costs
The comparatively lower unit costs enable the Good to reduce his prices right after his shift
in program focus and from that, he is constantly able to set lower prices than his competitors (see
Figure 7). As the Bad is constantly losing market shares, while missing his market goal to a
greater extent, he decreases his cost-plus margin until his prices nearly meet the costs per unit.
Thus, the Bad’s accumulated profits are leveling-off contrary to the Good (see Figure 8).
Cost per unit Accumulated profits
40 6M
38.12
36.25
€/Unit
34,37
32.5
0 Year 1 Year2Year3 Year 4 Years 9
The Good 0 Year1 Year2—Year3. = Year4—— Year 5
‘The Mediocre Time (Days) Time (Days)
The Good
The Bad
Figure 8: Cost per unit and accumulated profits
The Bad = =
However, the Good does not dominate the Bad on all performance figures, as can be seen in
Figure 9 on the aspect of time. The reasons for the Good's decreasing reputation in delivery de-
pendability are twofold: first, he is less engaged in programs on process acceleration, and second,
he builds up labor force delayed to increases in customer demand. The latter is due to the way he
adjusts his desired gross production rate. In the model, this is done by exponential smoothing of
the customer demand. This delaying effect could be partly eliminated by the use of a trend-
forecasting function, as discussed for example in Lyneis (1988).
Perceived delivery dependability relative to competitor Backlog
41 1,200
1.05 4,100
1
1,000
Dmnt
0.95 900
09 800
0 Year 1 Year 2 Year3-Year4 Year 5 0 Year 1 Year2 — Year3 Year 4 Year 5
The Good
The Good
The Mediocre Time (2yS) The Mediocre
TheBad =
The Bad ==
Figure 9: Perceived performance in time and orders backlog
Time (Days)
Even though the Bad is focusing on programs on process acceleration, he can hardly fulfill
customers’ expectations on delivery dependability. The slight increase in the Bad’s delivery de-
pendability from the middle of year 2 on is partly caused by the decreasing customer demand.
The fact that the Bad does not even perform well on the programs he is focusing on can be ex-
plained by the interplay between the sectors ‘management’, ‘human resources’, ‘improvement
programs’, and ‘manufacturing system’, as discussed earlier in this article (cf. Figure 2). As the
Bad’s improvement initiatives are failing to show progress to both managers and workers, the
program commitment decreases, as can be seen in Figure 10. The most influential factors for this
decay are the decreasing support by management and the increasing financial effort for process
improvement (see Figure 11). The latter affects the program commitment of the management and
can be traced back to the accumulated profits leveling-off in comparison to the steady financial
efforts for process improvements.
10
Program Commitment
— | Workers’ program commitment
. The Good
07 The Mediocre «
= The Bad
E Program commitment management
G 06 The Good
The Mediocre
0.5 The Bad
04
0 Year 1 Year 2 Year 3 Year 4 Year 5
Time (Days)
Figure 10: Program commitment of both, workers and management
In the case of the Good, the effect of the financial effort on program commitment is first ex-
ceeding the equilibrium run for approx. 2 years, then is falling below for another year, and is ex-
ceeding again until the end of the simulation run, which can not be clearly seen in Figure 11 due
to the scaling. Hence, the workers’ program commitment stays approx. at the same level as in the
equilibrium run.
Effect of perceived financial effort for process improvement
Perceived management support on management's program commitment
02 aE
0.25 0.85
E 03 & 07
0.35 0.55
04
o Year? Year Year Year Year 5
° Yeart Year2 Year3—— Ye 5
The Good ——— $$$
ard Year
Time (Days) The Good ‘Time (Days)
The Mediocre
The Bad ——
The Mediocre:
The Bad —
Figure 11: Effect of management support and financial improvement
effort on program commitment
Due to the vanishing commitment in the case of the Bad, the organizational learning effect
does not set in. Therefore, the workers do not benefit from the free time provided by the man-
agement, as can be seen for workers’ experiences with improvements in Figure 12. In the model
it is assumed that experiences in process improvements are more effective than formal training.
This means that a lag in experience can only be partly compensated by formal training. Hence,
the Bad’s workers cannot provide the same effort to process improvement even though they ex-
hibit a higher level in formal training. In the Good’s case, the decreasing and subsequently rising
behavior of the experience and training levels are due to the hiring of work force, as recruited
workers have to be trained by their experienced colleagues.
11
Average workers’ experiences with improvements Average workers’ training level
92 27
26.75
3 ts
5 4
2 : 26.5
2 2 26.25
26
a Yiwl Youu Vaau Wow Yous ° Year Year. Year Year4__ Year S
Time (Days) The Good Time (Days)
The Mediocre .
The Bad =
Figure 12: Workers’ experience and training level
Similar to the plot in Leone’s (1966) westem, the Good wins and the Bad loses in the end.
This is mainly due to the fact that the Bad fails to reduce unit costs while undertaking his im-
provement efforts. Hence both managers and workers lose their faith on the process improvement
programs, and from that they omit organizational leaning. These dynamic stories are under-
pinned by the empirical findings by Filippini et a/. (2001) and Vargas et al. (1999). However, it
should be emphasized that the way of reading and telling dynamic stories from the simulation
runs is indeed useful for both the modeler and reader of an article to gain confidence in the valid-
ity of the model. A modeler should not be satisfied that the model shows an expected behavior,
e.g. that model behavior fits to empirical data. By reading such dynamic stories from model be-
havior the modeler is forced to trace back its causes and by doing so he can discover anomalies in
his model. An anomaly exists if the modeler fails to trace back the causes for a certain behavior
pattern or if the pattern fails to fit in the dynamic plot the modeler is telling to the reader of an
article.
Conclusion and outlook
It could be shown that one can read insightful stories from the behavior of dynamic models and
that such dynamic story telling can increase the confidence in the model of both the modeler and
the reader of an article. The plot outlined on success and failure in process improvement pro-
grams tells us that plants should focus on programs that exhibit higher organizational complexity
rather than technical complexity. This is the case as the hard approach fails to contribute to or-
ganizational learning, and hence, organizational learning fails to gain momentum.
It should be noted that the different improvement patterns tested on the model so far are
comparatively simple. In reality one would expect to find several shifts of a plant’s foci and not
just one. Howsoever, it is easier to trace back model behavior with comparatively simple im-
provement pattems and it has been shown that this dynamic story reading contributes to gain con-
fidence in model validity. In a subsequent step, the model will be examined with more complex
patterns of process improvement.
12
References
Amnstrong, M. 2003. 4 Handbook of Human Resource Management Practice. 9th ed. London:
Sterling, VA : Kogan Page, http://www.loc.gov/catdir/toc/ecip047/2003016728.html
Carrol, JS, JD Sterman, and AA Marcus. 1997. Playing the Maintenance Game: How Mental
Models Drive Organizational Decisions. In Debating Rationality — Nonrational Aspects of
Organizational Decision Making, edited by J. J. Helper and R. N. Stern. Ithaca, N.Y .: ILR
Press.
Coyle, RG. 2000. Qualitative and Quantitative Modelling in System Dynamics - Some Research
Questions. System Dynamics Review 16 (3):225-144.
Filippini, R, A Vinelli, and C Voss. 2001. Paths of Improvement in Plant Operations. In High
Performance Manufacturing — Global Perspectives, edited by R. G. Schoeder and B. B.
Flynn. New Y ork (USA): John Wiley & Sons.
Ford, DN. 1999. A Behavioral Approach to Feedback Loop Dominance Analysis. System Dy-
namics Review 15 (1):3-36.
Forrester, JW. 1961. Industrial Dynamics. 6 ed. Cambridge (Ma.): MIT Press.
———. 1968. Market Growth as Influenced by Capital Investment. Industrial Management Re-
view 9 (2):83-105.
Gimeralp, B. 2006. Towards Coherent Loop Dominance Analysis: Progress in Eigenvalue Elas-
ticity Analysis. System Dynamics Review 22 (3):263-289.
Hanson, W. 1992. The Dynamics of Cost-Plus Pricing. Managerial and Decision Economics 13
(2):149-161.
Hill, T. 2000. Manufacturing Strategy — Text and Cases. 2 ed. Basingstoke, Hampshire: Palgrave.
Hines, J. 2005. Molecules of Structure Version 2.02 — Building Blocks for System Dynamics
Models. Lap Tec & Ventana Systems.
Hopp, WJ, and ML Spearman. 2001. Factory Physics — Foundations of Manufacturing Manage-
ment. 2nd ed. Boston: Irwin/McGraw-Hill.
Imai, M. 1986. Kaizen — The Key to Japan's Competitive Success. New Y ork: Random House
Business Division.
Ishikawa, K. 1985. What is Total Quality Control? The Japanese Way. Englewood Cliffs, N.J.:
Prentice-Hall.
Kampmann, CE, and R Oliva. 2006. Loop Eigenvalue Elasticity Analysis: Three Case Studies.
System Dynamics Review 22 (2):141-162.
Keating, EK, R Oliva, NP Repenning, S Rockart, and JD Sterman. 1999. Overcoming the Im-
provement Paradox. European Management Journal 17 (2):120-134.
Kim, DH. 1993. A Framework and Methodology for Linking Indiviual and Organizational Learn-
ing - Applications in TQM and Product Development, System Dynamics Group, Massachu-
setts Institute of Technology, Boston, Massachusetts.
13
Lane, DC, and C Smart. 1995. Reinterpreting "Generic Structure" - Evolution, Application and
Limitations of a Concept. System Dynamics Review 12 (2):87-120.
Leone, S. 1966. The Good, the Bad and the Ugly. In ‘Dollars’ Trilogy. Italy.
Liehr, M. 2001. Integrating Models of System Archetypes into a System Dynamics Platform -
Limits and Opportunities. In The 19th International Conference of the System Dynamics So-
ciety, July 23-27, at Atlanta, Georgia.
———. 2004. Komponentenbasierte Systemmodellierung und Systemanalyse: Erweiterung des
System-Dynamics-Ansatzes zur Nutzung im strategischen Management. Wiesbaden: Gabler
Edition Wissenschaft.
Lyneis, JM. 1988. Corporate Planning and Policy Design: A System Dynamics Approach, Cam-
bridge: MIT Press.
Maier, FH. 2000. Feedback Structures Driving Success and Failure of Preventive Maintenance
Programs. In Proceedings of the 7" International Conference of the European Operations
Management Association: Operations Management: Crossing Borders and Boundaries: The
Changing Role of Operations, at V lerick Leuven Ghent Management School, Belgium.
— ——. 2004. Verbesserungsinitiativen im Lichte von Komplexitat und Dynamik - Anmerkun-
gen zu einer Theorie der Erfolgsfaktoren von Verbesserungsprogrammen. In Komplexitdt
und Dynamik als Herausforderung fiir das Management, edited by F. H. Maier. Wiesbaden:
Deutscher Universitatsverlag.
Meadows, DH. 1982. Whole Earth Models and Systems. CoEvolution Quarterly 34 (Sum-
mer):98- 108.
Meyer, JP, and NJ Allen. 1991. A Three-component Conceptualization of Oorganizational Com-
mitment. Human Resource Management Review 1 (1):61-89.
Meyer, JP, and L Herscovitch. 2001. Commitment in the Workplace - Toward a General Model.
Human Resource Management Review 11 (3):299-326.
Milling, PM. 1972. First order Systems and S-shaped Growth: Materials for the NATO Advanced
Study Inst. in System Dynamics: NATO.
———. 1974. Der technische Fortschritt beim Produktionsprozess: Ein dynamisches Modell fiir
innovative Industrieunternehmen. Wiesbaden: Betriebswirtschaftlicher V erlag Gabler.
Mojtahedzadeh, M, D Andersen, and GP Richardson. 2004. Using Digest to implement the path-
way participation method for detecting influential system structure. System Dynamics Review
20 (1):1-20.
Nakajima, S. 1988. Introduction to TPM: Total Productive Maintenance. Cambridge, Mass.:
Productivity Press.
Neubert, MJ, and SH Cady. 2001. Program Commitment - A Multi-study Longitudinal Field
Investigation of Its Impact and Antecedents. Personnel Psychology 54 (2):421-448.
Oliva, R, S Rockart, and JD Sterman. 1993. Managing Multiple Improvement Efforts: Lessons
form a Semiconducter Manufacturing Site. In Advances in the Management of Organiza-
tional Quality, edited by D. B. Fedor and S. Ghosh. Stamford (CN.)/London (England): JAI
Press Inc.
14
Paich, M. 1985. Generic Structures. System Dynamics Review 1 (1):126-132.
Repenning, NP. 2002. A Simulation-Based Approach to Understanding the Dynamics of Innova-
tion Implementation. Organization Science 13 (2):109-127.
Repenning, NP, and JD Sterman. 1994. Unanticipated Side Effects of Successful Quality Pro-
grams: Technical Documentation. Massachusetts Institute of Technology.
———. 1997. Getting Quality the Old-Fashioned Way : Self-Confirming A ttributions in the Dy-
namics of Process Improvement. Sloan School of Management, MIT, Cambridge MA: Sloan
School of Management Massachusetts Institute of Technology.
—-——. 2001. Nobody Ever Gets Credit for Fixing Problems that Never Happened: Creating and
Sustaining Process Improvement. California Management Review 43 (4):64-88.
Richardson, GP. 1995. Loop Polarity, Loop Dominance, and the Concept of Dominant Polarity.
System Dynamics Review 11 (1):67-88.
Schneiderman, AM. 1988. Setting Quality Goals: Use Observed Rates of Continuous Improve-
ment to Position Targets. Quality Progress:51-57.
-—-- The Half-life/Complexity Matrix 1999. Available from
http: //}www.schneiderman.com/C oncepts/Half-life/half-life complexity_matrix.htm.
Senge, PM. 1994. The Fifth Discipline: The Art & Practice of the Learning Organization. New
Y ork et al.: Currency Doubleday.
———. 1999. The Dance of Change — The Challenges of Sustaining Momentum in Learning
Organizations. 1st ed. New Y ork: Currency/Doubleday.
Shiba, S, A Graham, and D Walden. 1993. 4 New American TOM — Four Practical Revolutions
in Management. Cambridge (Mass.), Norwalk (Conn.): Productivity Press.
Stata, R. 1989. Organizational Leaming - The Key to Management Innovation. Sloan Manage-
ment Review 30 (3):63-74.
Sterman, JD. 2000. Business Dynamics: Systems Thinking and Modeling for a Complex World.
Boston et al.
Sterman, JD, F Kofman, and NP Repenning. 1997. Unanticipated Side Effects of Successful
Quality Programs: Exploring a Paradox of Organizational Improvement. Management Sci-
ence 43 (4):503-521.
Thun, J-H. 2006. Maintaining Preventive Maintenance and Maintenance Prevention: Analysing
the Dynamic Implications of Total Productive Maintenance. System Dynamics Review 22
(2):163-179.
Vargas, G, and L Cardenas. 1999. 'Hard Practicies', 'Soft Practicies' and Business Management in
Manufacturing Firms. In 6” EUROMA Annual Conference: Managing Operations Networks,
at Venice, Italy.
15
According to Schneiderman (1988, 1999), a level of defects can be calculated at a particular
time ¢ with
¥ = You, = (Yan) €XP(-O(t~f,)) and p= 22),
HL
dY Y—Yoin _
— =—G(Y) —Yonin) €XP(—A(t — tg)
which can be transformed to 4
and thus the rate of improvement can be calculated as
where Yinin equals the minimum defect level achievable “hotly ¥,, equals the initial de-
fect level, ¢ equals time, t, equals initial time, and 7, equals the defect half life. Transformed
into integral equations, one receives
In(2)
HLi
imp, = (Y,-Ynin )* a, * 8 as improvement rate (supplemented with management focus
In(2) ,.
Ei
and Y, =, +f (der —imp;) dt for the level of defects i.
a; and workers’ effort f), det;=
(Yona, —¥;) as deterioration rate,
The different management foci for process improvement initiatives, are therefore set to
In(2)
“am 1)
The Good, the Bad and the Mediocre:
Creating Insightful Stories on Process Improvement
Supplementary file
Model equations
perceived delivery dependability=
SMOOTHi((throughput time competitors)/throughput time,ST market
,INI perceived delivery dependability)
a Dmal
perceived quality level=
SMOOTHi(quality level/(quality level competitors),ST market
,INI perceived quality level)
a Dmal
perceived price level=
SMOOTHi(price/(price level competitors), ST market,INI price level
)
i Dmal
management's program commitment=
SMOOTH((E perceived financial effort on commitment+E improvement results on commitment\
)/2,ST management's commitment
)
i Dmal
Perceived financial effort for process improvement=
XIDZ(expenses for process improvement, (revenues-expenses), expenses for process improvement\
/1e-012)
i Dmal
E perceived financial effort on commitment=
T E perceived financial effort on commitment(Perceived financial effort for process improvement\
)
i Dmal
T E perceived financial effort on commitment(
{(-1,0)-(1,1)],C1,0),(0,0),(0,1),(0.1,1),(0.2,0.75),(0.3,0.5),(0.4,0.25),(0.5,0.125\
),(0.6,0.075),(0.7,0.0375),(0.8,0.01875),(0.9,0))
expenses for process improvement=
budget for improvements/days per year
~ €/Day
ST management's commitment=
360
INI price level=
at
= Dmal
e |
budget for improvements= GAME (
normal budget for improvements)
e |
accumulated profits=INTEG (
revenues-expenses-expenses for process improvement- expenses for process improvement,
INI finanzielle mittel)
€
~ ~ :SUPPLEMENTARY
|
INI fraction of accumulated profits for process improvements=
0.1
~ Dmal
normal budget for improvements= INITIAL(
INI finanzielle mittel*INI fraction of accumulated profits for process improvements)
~ €
expected results processes’ quality=
workers’ effort in process improvement*LN(2)*INI likelihood of defects introduction/\
“expected half-life time"
~ Dmnl/Day
expected results quality control=
workers’ effort in process improvement*LN(2)*INI quality control/"expected half-life time"
~ Dmnl/Day
expected results processing time=
workers’ effort in process improvement*LN(2)*INI processing time/"expected half-life time"
a Dmal
T E improvement results on commitment(
[(0,0)-(2,1.5)],(0,0.25),(0.5,0.5),(1,1),(1.5,1.25),(2,1.5))
a Dmal
average salaries=
2500
~ €/(W orker* Month)
perceived improvement results=
(perceived results workers’ productivity +perceived results machinery down time+perceived results
processing time\
+perceived results suppliers’ quality +perceived results processes’ quality +perceived results
quality control\
6
e |
"max. likelihood of defects introduction"=
0.4
= Dmal
e |
expected results machinery down time=
workers’ effort in process improvement*LN(2)*INI machinery down time/"expected half-life time"
~ Dmnl/Day
"min. processing time"=
= Day
expected results workers’ productivity=
workers’ effort in process improvement* LN(2)*INI worker productivity/"expected half-life time"
~ Dmnl/Day
~ |
“expected half-life time"=
200
~ Day
perceived results workers' productivity=
gains in worker productivity/expected results workers’ productivity
~ Dmal
expected results suppliers’ quality=
workers’ effort in process improvement* LN(2)*INI fraction of defects from suppliers/\
“expected half-life time"
~ Dmnl/Day
perceived results machinery down time=
improvements in machinery down time/expected results machinery down time
a Dmal
perceived results processing time=
improvements in processing time/expected results processing time
a Dmal
perceived results quality control=
improvements in quality control/expected results quality control
a Dmal
E improvement results on commitment=
T E improvement results on commitment(perceived improvement results)
a Dmal
perceived results processes’ quality=
improvements in likelihood of defects introduction/expected results processes’ quality
a Dmal
e |
perceived results suppliers' quality=
improvements in suppliers’ quality/expected results suppliers’ quality
= Dmal
e |
captial unit costs=
(KAPITALKOSTEN +perceived inventory tumover* INTERNER ZINSSATZ* material costs)/net
production rate
~ €/Unit
e |
unit costs=
(unit labor costs+material costs+captial unit costs)
~ €/Unit
unit labor costs=
average salaries/days per month* work force/net production rate
~ €/Unit
focus machinery down time= GAME (
INI effort for machinery down time/SUM INI improvement effort)
~ Dmal
focus quality control=GA ME (
INI effort quality control/SUM INI improvement effort)
~ Dmal
gains in worker productivity=
("max. worker productivity"-worker productivity)* LN(2)*workers' effort in process improvement\
* effort worker productivity
/“half-life time worker productivity"
~ Unit/(W orker*Day* Day)
|
improvements in processing time=
(processing time-"min. processing time")*LN( 2 )/"half-life time processing time"*workers' effort in
process improvement
* effort processing time
~ Day/Day
improvements in likelihood of defects introduction=
(likelihood of defects introduction-"min. likelihood of defects introduction")*LN(2)\
/"half-life time likelihood of defects introduction"
* effort for likelihood of defects introduction* workers’ effort in process improvement
~ Dmnl/Day
improvements in suppliers' quality=
(fraction of defects from suppliers-"min. fraction of defects from suppliers"
)/"half-life time fraction of defects from suppliers"* LN(2)*workers' effort in process improvement
*effort fraction of defects from suppliers
~ Dmnl/Day
improvements in machinery down time=
LN(2)*(machinery down time-"min. machinery down time")/"half-life time machinery down time"
*effort for machinery down time
*workers' effort in process improvement
~ Dmnl/Day
improvements in quality control=
("max. quality control"-quality control)*LN(2)*workers' effort in process improvement
*effort quality control
/“half-life time quality control"
~ Dmnl/Day
effort for machinery down time=
focus machinery down time*SUM INI improvement effort
= Dmal
effort quality control=
focus quality control*SUM INI improvement effort
~ Dmal
focus processing time= GAME (
INI effort processing time/SUM INI improvement effort)
~ Dmal
focus worker productivity=GAME (
INI effort worker productivity/SUM INI improvement effort)
~ Dmal
focus likelihood of defects introduction= GAME (
INI effort for likelihood of defects introduction/SUM INI improvement effort)
~ Dmal
effort for likelihood of defects introduction=
focus likelihood of defects introduction* SUM INI improvement effort
a Dmal
focus on fraction of defects from suppliers= GA ME (
INI effort fraction of defects from suppliers/SUM INI improvement effort)
a Dmal
effort worker productivity=
focus worker productivity*SUM INI improvement effort
a Dmal
effort processing time=
focus processing time*SUM INI improvement effort
a Dmal
effort fraction of defects from suppliers=
focus on fraction of defects from suppliers*SUM INI improvement effort
a Dmal
likelihood of defects introduction= INTEG (
deteriorations in likelihood of defects introduction-improvements in likelihood of defects introduction\
INI likelihood of defects introduction)
= Dmal
e |
"max. processing time"=
= Day
e |
quality control=INTEG (
-deteriorations in quality control+improvements in quality control,
INI quality control)
~ Dmal
deteroations in suppliers' quality=
("max. fraction of defects from suppliers"-fraction of defects from suppliers
)*LN(2)/ET fraction of defects from suppliers
~ Dmnl/Day
"max. machinery down time"=
0.2
~ Dmal
SUM INI improvement effort= INITIAL(
INI effort worker productivity +INI effort for likelihood of defects introductiont+INI effort processing
time\
+INI effort fraction of defects from suppliers NI effort for machinery down time+INI effort
quality control\
)
~ Dmal
fraction of defects from suppliers= INTEG (
deteroations in suppliers’ quality-improvements in suppliers' quality,
INI fraction of defects from suppliers)
a Dmal
machinery down time= INTEG (
-improvements in machinery down time+deteriorations in machinery down time,
INI machinery down time)
a Dmal
worker productivity= INTEG (
-loses in worker productivity +gains in worker productivity,
INI worker productivity)
~ Unit/(W orker*Day)
INI effort for likelihood of defects introduction= INITIAL(
(LN(2)/ET likelihood of defects introduction)*("max. likelihood of defects introduction"\
-INI likelihood of defects introduction)/(workers' effort in process improvement
*(LN(2)/"half-life time likelihood of defects introduction")*(INI likelihood of defects introduction\
-"min. likelihood of defects introduction"
= Dmal
e |
INI effort processing time=INITIAL(
(LN(2)/ET processing time)*("max. processing time"-INI processing time)/(workers' effort in process
improvement\
*(LN(2)/"half-life time processing time")* (INI processing time-"min. processing time"
))
= Dmal
e |
INI effort fraction of defects from suppliers= INITIAL(
(LN(2)/ET fraction of defects from suppliers)*("max. fraction of defects from suppliers"\
-INI fraction of defects from suppliers
)(workers' effort in process improvement*(LN(2)/"half-life time fraction of defects from suppliers"\
)*(INI fraction of defects from suppliers
-"min. fraction of defects from suppliers"
))
~ Dmal
INI effort for machinery down time= INITIAL(
(LN(2)/ET machinery down time)*("max. machinery down time"-INI machinery down time
)(workers' effort in process improvement*(LN(2)/"half-life time machinery down time"\
)*(INI machinery down time
-"min. machinery down time"
))
~ Dmal
INI effort quality control= INITIAL(
(INI quality control-MINIMALER ANTEIL FEHLERENTDECKUNG)*LN(2)/ET quality
control/((""max. quality control"\
-INI quality control)*LN(2)*workers' effort in process improvement
/"half-life time quality control"))
~ Dmal
deteriorations in likelihood of defects introduction=
("max. likelihood of defects introduction"-likelihood of defects introduction)*LN(2)\
/ET likelihood of defects introduction
~ Dmnl/Day
deteriorations in machinery down time=
("max. machinery down time"-machinery down time)*LN(2)/ET machinery down time
~ Dmnl/Day
"max. fraction of defects from suppliers"=
0.4
a Dmal
GAME INTERVAL=
GAME(30)
on Day
ns on :SUPPLEMENTARY
|
program commitment workers=INTEG (
(change in workers' commitment+E management on workers' commitment),
INI workers' commitment)
= Dmal
e |
"min. worker productivity"=
~ Unit/(Day*W orker)
MINIMALER ANTEIL FEHLERENTDECKUNG=
0.8
= Dmal
e |
processing time=INTEG (
deteriorations in processing time-improvements in processing time,
INI processing time)
~ Day
deteriorations in quality control=
(quality control-MINIMALER ANTEIL FEHLERENTDECKUNG)*LN(2)/ET quality control
~ Dmnl/Day
loses in worker productivity=
(worker productivity-"min. worker productivity")*LN(2)/ET worker productivity
~ Unit/(W orker* Day* Day)
|
ET machinery down time=
1080
~ Day
INI effort worker productivity=INITIAL(
(INI worker productivity-"min. worker productivity")*LN(2)/ET worker productivity/((\
"max. worker productivity"
-INI worker productivity)*LN(2)*workers' effort in process improvement
/"half-life time worker productivity"))
a Dmal
deteriorations in processing time=
(max. processing time"-processing time)*LN(2)/ET processing time
~ Day/Day
INI workers' commitment= INITIAL(
0.786821)
a Dmal
machinery capacity=
10000*(1-machinery down time)
~ Unit/Day
T E market share on margin(
{(0,0)-(2,2)],(2,2),(1.3, ay
(1.2,1.75),(1.1,1.5),(1.05,1.1),(1,1),(0.95,0.9),(0.9,0.5\
),(0.8,0.25),(0.7,0),(
,0))
= Dmal
e |
E market share on margin=
T E market share on margin(market share/desired market share)
= Dmal
e |
price=
SMOOTH(unit costs*(1+desired margin*E market share on margin), AT price)
~ €/Unit
e |
desired market share=
INI traditional market share
= Dmal
decline in memmories in lay offs=
perceived job security/ST forgetting lay offs
~ Dmnl/(Day* Day)
|
willingness to hire=
GAME(1)
~ Dmal
INI experience level=
"averag. experiences new recruits"*hiring+ on the job experiences/(1/DT forgetting time experiences\
+1/work force* (laying off+fluctuation+on the job experiences/"max. averag. experience level''\
)
~ Hour
fraction of workers' productivity for training=
(1-training effort-fraction of training effort)
~ Dmal
INI workers' productivity=
10
~ Unit/(W orker*Day)
~ |
INI perceived quality level=
1
a Dmal
~ |
INI perceived delivery dependability=
1
a Dmal
fraction of training effort=
"max. fraction working day for improvements"*program commitment workers
a Dmal
INI training level=
"averag. training new recruits"*hiring+intensity of training*work force/(1/DT forgetting time training\
+1/work force* (fluctuation Haying off)+intensity of training/"max. averag. training level"\
)
s Hour
e |
workers’ effort in process improvement=
"averag. improvement capabilities worker"*program commitment workers
= Dmal
e |
gains in experience through hiring=
"averag. experiences new recruits"*hiring
~ Hour/Day
e |
ST workers’ productivity=
30
~ Day
DT forgetting time experiences=
1800
~ Day
~ |
work day=
8
~ Hour/(Day*W orker)
~ |
ordering=
customer order rate
~ Unit/Day
E commitment on gains in commitment=
T E commitment on gains in commitment(program commitment workers)
~ Dmal
~ |
expenses=
unit costs*net production rate
~ €/Day
~ |
desired management support per worker=
~ Hour/W orker/Day
desired management support=
work force*desired management support per worker*program commitment workers
~ Hour/Day
T effekt arbeitsplatzsicherheit auf commitment(
[(0,-1)-(0.1,0)],(0,0),(0.0075,-0.35),(0.02,-0.6),(0.035,-0.825),(0.06,-0.95),(0.07,\
-0.975),(0.09,-1),(0.1,-1))
a Dmal
T E commitment on gains in commitment(
[(0,0)-(1,1)],(0,0),(0.25,0.5),(0.3,0.58),(0.35,0.65),(0.42,0.72),(0.5,0.75),(0.58,0.72\
),(0.64,0.67),(0.69,0.6),(0.75,0.5),(1,0))
s Dmal
e |
T E price on market share(
[(0.75,0)-(1.25,1.6)],(0.75,1.6),(0.8,1.4),(0.85,1.25),(0.9,1.15),(0.95,1.05),(1,1),\
(1.05,0.95),(1.1,0.85),(1.15,0.75),(1.2,0.6),(1.25,0.4))
= Dmal
e |
gross production rate=
min(production capacity, WIP/processing time)
= Unit/Day
e |
fluctuation=
work force/DT membership
~ Worker/Day
~ |
market share=
E quality on market share*E price on market share*E delivery dependability on market share\
*traditional market share
~ Dmal
training effort=
intensity of training*work day
~ Dmal
throughput time=
backlog/net production rate
~ Day
throughput time competitors= INITIAL(
throughput time*INI perceived delivery dependability)
on Day
“average. lay offs"=
laying off/work force
~ Dmnl/Day
DT membership=
7200
on Day
“averag. improvement capabilities worker"=
E experiences on improvement capabities* "averag. training level"/"max. averag. training level"
+E training on improvement capabities
*"averag. experiences"/"max. averag. experience level"
a Dmal
“averag. experiences" =
workers' experiences/work force
a Hour/W orker
e |
“averag. experiences new recruits"=
0.1
= Hour/Worker
e |
“averag. training level"=
workers’ training level/work force
= Hour/W orker
e |
“averag. training new recruits"=
0.1
= Hour/W orker
e |
E delivery dependability on market share=
T E delivery dependability on market share(perceived delivery dependability)
~ Dmal
E management support=
T E management support(desired management support/management support)
~ Dmal
E quality on market share=
T E quality on market share(perceived quality level)
~ Dmal
E perceived job security on commitment=
T effekt arbeitsplatzsicherheit auf commitment(perceived job security)
~ Dmal
perceived fluctuation=
SMOOTH (fluctuation, ST fluctuation )
~ Worker/Day
E experiences on gain in experiences=
MAX(1-"averag. experiences"/"max. averag. experience level" ,0)
a Dmal
E price on market share=
T E price on market share(perceived price level)
a Dmal
E training on gain in training=
MAX(1-"averag. training level'/"max. averag. training level",0)
a Dmal
E management on workers' commitment=
(management's program commitment-program commitment workers)/days per year
~ Dmnl/Day
E experiences on improvement capabities=
0.2
= Dmal
e |
E training on improvement capabities=
0.8
= Dmal
e |
increase in memmories in lay offs=
MAX(“average. lay offs'-perceived job security, 0 )/ST memmroies lay offs
s Dmnl/(Day* Day)
e |
hiring=
a‘ willingness to hire*MA X (Workerliicke/EINSTELLUNGV ERZOGERUNGSZEIT +perceived
fluctuation\
0)
~ Worker/Day
EINSTELLUNGVERZOGERUNGSZEIT=
90
~ Day
~ |
laying off=
willingness to lay off*MA X (Workerlitcke*(-1),0)/DT laying off
~ Worker/Day
willingness to lay off=
GAME(1)
~ Dmal
~ |
DT laying off=
90
on Day
on the job experiences=
work force* program commitment workers* work day*"max. fraction working day for improvements"
~ Hour/Day
workers’ experiences= INTEG (
gains in experience through hiring+gains in experience-loses in experience from fluctuation and laying
off\
-loses in experiences from forgetting,
INI experience level)
at Hour
loses in experience from fluctuation and laying off=
"averag. experiences"* (laying off+fluctuation)
~ Hour/Day
loses in experiences from forgetting=
workers’ experiences/DT forgetting time experiences
~ Hour/Day
gains in experience=
E experiences on gain in experiences*on the job experiences
~ Hour/Day
perceived job security= INTEG (
increase in memmories in lay offs-decline in memmories in lay offs,
laying off/work force)
~ Dmnl/Day
e |
DT forgetting time training=
1800
~ Day
perceived inventory turnover=
SMOOTH(inventory turnover, ST perceived inventory tumover)
~ Day
work force=INTEG (
hiring-laying off-fluctuation,
desired work force)
~ Worker
intensity of training=
management's training goal*"max. averag. training level"/days per year
~ Hour/(Day*W orker)
|
T E delivery dependability on market share(
[(0,0)-(2,1)],(0,0.4),(0.3,0.425),(0.575,0.525),(0.8,0.7),(0.9,0.9),(1,1),(2,1))
~ Dmal
T E management support(
{(0,-1)-(2,1)],(0,1),(1,1),(1.05,0.825),(1.1,0.025),(1.25,-0.55),(1.35,-0.8),(1.5,-0.95\
),(2,-1))
at Dmal
T E quality on market share(
{(0.75,0)-(1.25,2)],(1.25,1.6),(1.2,1.4),(1.15,1.25),(1.1,1.15),(1.05,1.05),(1,1),(0.95\
,0.95),(0.9,0.85),(0.85,0.75),(0.8,0.6),(0.75,0.4))
a Dmal
ST memmroies lay offs=
1
on Day
desired production rate=
desired throughput/perceived process yield
~ Unit/Day
ST forgetting lay offs=
1800
s Day
e |
days per year=
360
= Day
e |
traditional market share=
SMOOTHi(market share,ST market share, INI traditional market share)
= Dmal
e |
ST fluctuation=
14
~ Day
price level competitors= INITIAL(
INI price level* price)
~ €/Unit
production capacity=
min(machinery capacity, fraction of labor productivity for production)
~ Unit/Day
workers' training level= INTEG (
gains in training through hiring Hraining-loses in training from fluctuation and laying off\
-loses in training from forgetting,
INI training level)
~ Hour
gains in training through hiring=
hiring*"averag. training new recruits"
~ Hour/Day
loses in training from fluctuation and laying off=
"averag. training level"* (laying off+fluctuation)
~ Hour/Day
loses in training from forgetting=
workers’ training level/DT forgetting time training
~ Hour/Day
fraction of labor productivity for production=
work force* worker productivity* (1-training effort-fraction of training effort)
~ Unit/Day
customer order rate=
market demand* market share
~ Unit/Day
WOM=
E commitment on gains in commitment*(E management support+E perceived job security on
commitment\
+E improvement results on commitment)/3
= Dmal
e |
ST market share=
360
= Day
e |
desired work force=
desired gross production rate/Perceived W orkers' Productivity/fraction of workers’ productivity for
training
= Worker
change in workers' commitment=
WOM/T communication
~ Dmnl/Day
training=
intensity of training*work force*E training on gain in training
~ Hour/Day
Workerliicke=
desired work force-work force
~ Worker
Perceived W orkers' Productivity=
SMOOTHi(worker productivity, ST workers' productivity,INI workers' productivity)
~ Unit/(Day*W orker)
T communication=
30
at Day
~ |
INI traditional market share=
0.1
a Dmal
"max. averag. training level"=
40
a Hour/W orker
management's training goal=
GAME(0.5)
a Dmal
quality level=
1-fraction of defects in supply
a Dmal
"max. fraction working day for improvements"=
GAME(0.1)
= Dmal
e |
"Max. averag. experience level"=
100
= Hour/W orker
e |
management support=
desired management support per worker*work force*management's program commitment
~ Hour/Day
quality level competitors= INITIAL(
INI perceived quality level* quality level)
~ Dmal
INI perceived process yield=
1-quality control* (likelihood of defects introduction-likelihood of defects introduction\
*fraction of defects from suppliers
+fraction of defects from suppliers)
~ Dmal
perceived process yield=
SMOOTHi( ratio net to gross production , ST process yield ,INI perceived process yield\
)
~ Dmal
INI finanzielle mittel=
1e+006
~ €
revenues=
net production rate*price
~ €/Day
ET processing time=
1080
on Day
ET quality control=
1080
on Day
ET worker productivity=
1080
on Day
ET fraction of defects from suppliers=
1080
on Day
desired margin=
GAME(0.15)
= Dmal
e |
KAPITALKOSTEN=
100
~ €
2 |
days per month=
30
~ Day/Month
deliveries=
DELAY 1(orders,DT deliveries)
~ Unit/Day
fraction of defects in supply=
ZIDZ(undetected defects,net production rate )
~ Dmal
market demand=
5000
~ Unit/Day
material costs=
10
~ €/Unit
INTERNER ZINSSATZ=
0.1
a Dmal
undetected defects=
defects in production-defects elimination
~ Unit/Day
on Day
desired WIP level=
desired production rate* processing time
a Einheit
defects elimination=
quality control* defects in production
~ Unit/Day
e |
INI quality control=
0.9
= Dmal
e |
WIP=INTEG (
+feeding in processes-gross production rate,
desired WIP level)
= Unit
e |
"max. quality control"=
1
~ Dmal
“half-life time quality control"=
150
~ Day
"max. worker productivity"=
12
~ Unit/(W orker*Day)
~ |
INI worker productivity=
10
~ Unit/(Day*W orker)
“half-life time worker productivity"=
350
~ Day
INI fraction of defects from suppliers=
0.2
a Dmal
~ |
INI processing time=
on Day
"min. fraction of defects from suppliers"=
0
a Dmal
"min. machinery down time"=
0
a Dmal
“half-life time machinery down time"=
= Day
e |
“half-life time processing time"=
400
= Day
e |
“half-life time fraction of defects from suppliers"=
700
= Day
e |
“half-life time likelihood of defects introduction"=
400
~ Day
"min. likelihood of defects introduction" =
0
~ Dmal
inventory tumnover=
inventory tumover W1P+inventory tumover materials
~ Day
INI likelihood of defects introduction=
0.2
~ Dmal
feeding in processes=
min(desired gross production rate,materials/setup time)
~ Unit/Day
ET likelihood of defects introduction=
1080
on Day
orders=
MAX (0, desired gross production rate+correction materials )
~ Unit/Day
ratio net to gross production=
ZIDZ(net production rate, gross production rate )
a Dmal
desired materials=
desired gross production rate* desired materials coverage
on Unit
inventory tumover materials=
ZIDZ(materials, feeding in processes)
on Day
e |
INI machinery down time=
0.1
= Dmal
= Day
inventory turnover WIP=
ZIDZ(WIP, net production rate )
= Day
e |
ST perceived inventory tumover=
~ Day
~ |
ST process yield=
7
~ Day
defects in materials= INTEG (
defects form supplier-defects feeding in processes,
fraction of defects from suppliers* desired materials)
~ Unit
defects in WIP= INTEG (
defects feeding in processes+defects introduction due to processes- defects in production\
feeding in processes* (likelihood of defects introduction-likelihood of defects introduction\
*fraction of defects in materials+fraction of defects in materials)*processing time\
)
~ Unit
defects introduction due to processes=
likelihood of defects introduction* (feeding in processes-defects feeding in processes\
)
a Unit/Day
~ |
delivering=
net production rate
~ Unit/Day
materials= INTEG (
+deliveries- feeding in processes,
desired materials)
on Unit
net production rate=
gross production rate-defects elimination
~ Unit/Day
e |
backlog=INTEG (
ordering- delivering,
desired throughput time* ordering)
= Unit
e |
desired materials coverage=
14
= Day
e |
correction time materials=
14
= Day
desired gross production rate=
MAX (0,desired production rate+correctionW IP)
~ Unit/Day
correction materials=
(desired materials-materials)/correction time materials
~ Unit/Day
~ |
desired throughput=
backlog/desired throughput time
~ Unit/Day
~ |
desired throughput time=
2
~ Day
correctionWIP=
(desired WIP level-WIP)/correction time WIP
~ Unit/Day
~ |
correction time WIP=
on Day
on Day
fraction of defects in WIP=
defects in WIP/WIP
a Dmal
defects in production=
gross production rate*fraction of defects in WIP
~ Unit/Day
defects form supplier=
deliveries*fraction of defects from suppliers
~ Unit/Day
e |
defects feeding in processes=
fraction of defects in materials*feeding in processes
~ Unit/Day
e |
fraction of defects in materials=
defects in materials/materials
= Dmal
e |
