A system dynamics computer-based learning environment
for the formulation of manufacturing strategy
Emmanuel D. Adamides
Laboratory of Industrial Management and Information Systems
Department of Mechanical Engi ‘ing and Aer i
University of Patras
Rion Campus, Patras 26500, Greece
adamides@mech.upatras.gr
Abstract
The paper discusses how the calibration and use of a Computer Based Learning
Environment (CBLE), which is based on a system dynamics model, can form the
basis of strategy formulation processes at the operations level. The rationale behind,
the structure and the elements of the SYDMAS CBLE, as well as its embedment in a
scenario-driven manufacturing strategy formulation process are presented. Through a
use case, it is shown how the CBLE can enhance the manufacturing strategy
formulation process by providing a dynamic perspective and by effectively supporting
the related social and knowledge processes.
1. Introduction
Over the last three decades, based on the realisation of manufacturing’s potential to
act as a direct source of competitive advantage (Skinner, 1969; Hayes and
Wheelwright, 1984), manufacturing strategy (otherwise synonymous to operations
strategy for more service-oriented firms) has emerged as one of the most important
constituent parts of corporate strategy. The formalisation of the development and
deployment processes of this functional strategy has been of interest to many
academics and practitioners. The majority of authors, influenced by the application of
the “rationalistic” paradigm of strategy, have proposed tools and procedures for
statically assessing the manufacturing’s internal and external environment at a
particular instance in time, and for identifying the actions needed to achieve fit among
them (e.g. Platts and Gregory, 1990; Mills, et al., 1995; Joseph, 1999; Quezada et al.,
1999; Hill, 2000; Tan and Platts, 2004). (It should be noted that currently
manufacturing strategy is assumed to include all activities along the value chain
through which physical objects (raw materials, components, products) are moving.)
Nevertheless, given the dynamic and unpredictable nature of environmental changes,
as well as the dynamic evolution of the internal resources and capabilities, purely
rationalist approaches to manufacturing strategy formulation seem to be insufficient.
Manufacturing-related resources and capabilities take time to build and the
opportunities identified may disappear, or change to a lesser of greater degree, in the
mean time (Hayes et al., 1996).
On a_ different to the “rationalist? and prescriptive perspective, the
emergent/evolutionary strategy paradigm (Mintzberg and Waters, 1985; Mintzberg
and Lampel, 1999) emphasizes retrospective sense-making of strategic initiatives but
undermines the fact that strategy-making, in one way or another, takes place in
anticipation of the future before actions are decided and implemented. Moreover, its
basic philosophy disempowers the role of managers and is quite complicated to apply
in real situations (Van der Heijden, 1996), especially in manufacturing/operations and
other functionally interdependent strategies.
To address the requirements of explicitly taking into account the dynamics of
manufacturing system in establishing a reliable formulation process, we have adopted
a processualist/learning perspective in strategy making. In this view, the strategy
process per se has more, or at least the same, importance as its content and outcome.
Towards this end, we have developed and used a novel formulation process which
relies on learning technology. Our process diverts from traditional scenario-building
exercises (Codet, 1987; Schoemaker, 1995; Van der Heijden, 1996) in that it relies on
the use of SYDMAS (System Dynamics in MAnufacturing Strategy), a Computer-
Based Learning Environment (CBLE) specific to the manufacturing/operations
strategy process. The technology used diverts from traditional decision-support tools
for the same task (e.g. TAPS of Tan and Platts (2004a; 2004b)) in that is fully
parameterised, operates at the purely strategic level, explicitly considers the dynamics
of the manufacturing system, and, more importantly, it does not fully rely on the
subjectively-developed “objective” knowledge of the developer. Instead, the
calibration of a pre-structured system dynamics simulation model and the execution of
simulations constitute learning exercises, where knowledge elicitation from diverse
sources and recombination take place. In the long term, this enhances team learning
and decision-making capability, and provides the medium for the development of a
shared perspective for managers with diverse backgrounds and responsibilities, who
are, however, stakeholders of manufacturing strategy. In other words, the model
serves as a “transitional object” for mental models (Papert, 1980; Morecroft, 2004).
This paper concentrates on the characteristics of the learning environment
(SYDMAS) and its embedment in the overall manufacturing strategy process. We
show how a CBLE can support a strategy process which satisfies the requirement for
a dynamic perspective in uncertain environments. In addition, we demonstrate the use
of SYDMAS and its associated process through an application example of
manufacturing strategy formulation in a real company. Finally, we discuss our
experiences from this application.
2. Coping with dynamics: A dynamic model of manufacturing
strategy
2.1 Conceptual issues
In a dynamic resource-based perspective (Dierickx and Cool, 1989), manufacturing
strategy can be thought as a sequence of decisions and actions on the accumulation
and combination of manufacturing tangible and intangible assets stocks (resources,
capabilities and competences), which are necessary for achieving a sustainable fit of
the firm with its environment (Grant, 1991; Slack and Lewis, 2002). Resources are
stocks of available factors which are owned or controlled by the firm, and may
include production equipment, planning and scheduling software, machine operators,
reputation for quality products etc. Capabilities, on the other hand, can be defined as
the capacity of a company to deploy resources, or combinations of resources using
organisational processes (Amit and Schoemaker, 1993), or routines (Nelson and
Winter, 1992). For instance, a company can use its flexible equipment (resources)
effectively if it has a capability in complex scheduling. The stock levels of capabilities
are accumulated through the execution of organisational activities and influence the
rates of resource accumulations (complex scheduling capability facilitates the
deployment of flexible machinery). Both capability and resource accumulations may
be self reinforcing, e.g., an existing capability in complex scheduling many be easily
extended horizontally by training internally new schedulers, or vertically by learning
more complex and more efficient methods. In addition, resource building activities
influence the rate of capability accumulation (frequent deployment of flexible
equipment in production processes increases the capability of developing complex
schedules).
The combination/architecture of manufacturing strategic assets and their stock levels
define not only the range and the economies of the activities in which the firm can
engage at any point in time (Ghemawat et al., 2001), but also plays a decisive role on
the choices of the future competitive objectives by determining the difficulty involved
in developing the newly required assets. Specific assets, at specific stock levels may
augment or limit the decision space of future manufacturing and corporate strategies
(path-dependent trade-offs). For instance, a firm that has invested in dedicated
capacity can easily adopt cost leadership strategies by exploiting its capacity and by
being supported by its infrastructural attributes (e.g. highly specialised automation,
untrained personnel, lengthy production schedules etc.) which have been tuned to
large-scale operations. In general, however, the same firm will have a difficulty in
adopting a mass-customisation strategy after developing structural and infrastructural
resources for mass production.
2.2 Manufacturing strategy dynamics
Integrating the above concepts within the production/operations context, we have
developed the resource-based model of manufacturing strategy shown in figure 1.
Being a complex system, the manufacturing activity-asset architecture exhibits
behaviours which are governed by the spatial and temporal interconnections among its
elements, that is, the way functional assets and processes (sets of activities) are related
over time. To conceptualise the roots of the dynamic behaviour of such a system, we
rely on the modelling language of system dynamics. In system dynamics, the
accumulation of assets as a result of the execution of specific activities over time can
be modelled by stocks, whereas the rates of accumulation (decisions, activities and
processes) and erosion/depletion as flows (Morecroft, 1999; Adamides, 2002;
Mollona, 2002; Warren, 2002; Groessler, 2005).
actrescoeff — caprescoetf ()
res obj coeff
mfg activities
act cap coeff ep inesement
cap out
res cap coeff
cap out coeff
Figure 1 The system dynamics model of manufacturing strategy
In the model of figure 1, mfg_resources and mfg_capabilities are two sets of stocks
(array stocks) representing the accumulation of manufacturing-related resources and
capabilities, respectively. On the other hand, res_in is a set of (array) flows indicating
the rate of accumulation of manufacturing resources, whereas cap_in is a set of flows
representing the rate of accumulation of manufacturing capabilities. Both rates depend
on the intensity and frequency of execution of a set of resource and capability
building activities (mfg_activities) which may belong to specific operational processes
(or routines), or may be intentional asset development decisions, such as investment
in equipment and facilities. The rates of resource and capability erosions are
represented by the set of flows res_out and cap_out, respectively, which depend on
the current level of their corresponding stock (res_out_coeff and cap_out_coeff
through the intermediate variables res_er_rate and cap_er_rate).
The performance of the manufacturing function is measured with respect to the stock
levels of its resources. Resource levels correspond to the values of the typical
manufacturing performance objectives of cost, flexibility, dependability, quality and
speed through a matrix (two-dimensional array) of coefficients res_obj_coeff. The
links between mfg_resources and cap_in on the one hand, and mfg_capabilities and
res_in on the other, indicate the mutual dependence of resources and capabilities.
How resources and capabilities are linked (whenever specific pairs are linked) is
indicated in the converter matrices of coefficients res_cap_coeff and cap_res_coeff of
appropriate dimensions. The matrices of coefficients act_res_coeff and act_cap_coeff
denote the relation (if it exists) of specific activities to the resources and capabilities
of the manufacturing function (through the intermediate variables res_increment and
cap_increment, respectively). The mutual effect of the stock levels of resources and
capabilities (facilitating or impeding) is indicated by the converter matrices
res_res_coeff and cap_cap_coeff within the corresponding reinforcing loops. They
indirectly indicate any trade-offs among resources and among capabilities, that is
whether the earlier development of a specific resource hinders the execution of
activities towards the development of trade-off manufacturing resources.
Consequently, what constitutes manufacturing strategy in the logic of the above
model, is the establishment of the conditions which create reinforcing loops among
specific sets of resources and capabilities. This becomes possible by identifying and
building distinct manufacturing resources, capabilities and/or linkages among them,
and/or by selecting improvement programs which result in faster, or superiorly timed,
resource and capability accumulations.
It should be noted that the purpose of the model of figure 1 is to demonstrate the basic
elements of the resource-capability system. Clearly, the actual working model used in
SYDMAS contains more elements for reasons imposed by the technicalities of the
simulation environment, for additional calculations (cost and risk), as well as for
implementing the graphical displays.
2.3 SYDMAS - A CBLE for the formulation of manufacturing strategy
Computer Based Learning Environments (CBLE) or Microworlds have been
developed and used for enhancing learning in strategic planning and decision making
processes (Morecroft, 1988; Issacks and Senge, 1994). CBLEs for strategic
management have been associated with system dynamics because at the heart of the
majority of CBLEs lies a flexible dynamic simulation model at this level of
abstraction (not discrete event) of the issue or the problem to be dealt with. Through
user-friendly interfaces, they provide the medium for simulation-supported scenario
building, experimentation and evaluation. According to Riis and Smeds (1998), what
constitutes a learning environment extends from a simple simulation model and its
interface to include the physical and social setting, the facilitation and the debriefing
of the simulation. In the manufacturing/operations area, the most frequent use of a
CBLE is for training purposes (learn by doing) (Smeds, 2003), while their embedment
in actual organisational tasks (do by learning), is principally associated to discrete
event process models for change management (development and re-engineering of
administrative and industrial processes, e.g., Taskinen, 2003).
In the manufacturing strategy context, the SYDMAS learning environment allows
strategists to address and experiment on questions such as: When the market
requirements and the specific manufacturing competence will be aligned? How long
will that take? When should the firm increase the effort to achieve this on time? Do
current decisions enhance or limit the long-term strategic flexibility of the
manufacturing function? What is the emerging performance of the manufacturing
competences?
The kernel of the SYDMAS computer-based learning environment is the
implementation of the resource-based model of manufacturing strategy of the
previous section in the system dynamics simulation environment ithink Analyst.
Users can calibrate and manipulate the model through friendly interfaces. They can
specify the current/initial state of resources and capabilities, as well as the
impact/contribution factors and the scaling coefficients of the model. The execution of
activities can be specified in two modes: either interactively during the simulation as
rates (e.g. monthly), or as commitments of a particular intensity that take place at a
specific time period. In the current version of the SYDMAS prototype, up to nine
activities with their associated costs and risk factors can be specified. The
performance sought can be defined in terms of the performance objectives of cost,
flexibility, dependability, quality and speed for up to five manufacturing processes
simultaneously which are directly related to products or product groups. Aggregates
for more than one product group can also be defined (the default mode of aggregation
is by averaging, but other modes can also be specified). Simulations can be executed
either by fixing the parameters of the model at the beginning, or interactively in the
management flight simulator mode. Absolute or comparative (discrepancies)
performance levels with respect to the required ones can be displayed. Total costs,
cost profiles and costs per activity, as well as risk estimates are also calculated and
displayed. Costs are defined per activity and are accumulated according to the
frequency and intensity of activity execution.
In the manufacturing strategy formulation process presented in the following section,
the use of SYDMAS serves a two-fold purpose. First, the determination of linkage
coefficients to calibrate the model provides the incentive to research, discuss and
make sense of the current state of the company’s manufacturing operations and
strategies. Secondly, the experimentation and evaluation of alternative activity
execution schemes and interventions on the linkages among the elements of the model
constitute a learning exercise that cannot be accomplished in its absence. In both
cases, the model acts a medium to engage managers in a strategic conversation of
immense depth and value for the company.
3. Coping with uncertainty: Manufacturing strategy as a learning
exercise
3.1 The overall process
The methodology developed is a facilitator-driven process that combines two of the
most widely used tools of the procesualist, or learning, school of strategy
development: construction of scenarios and group model building and evaluation. As
with other participative approaches to strategy formulation that use information
technology for mapping (e.g. the SODA methodology and Group Explorer (Eden and
Ackermann, 1998, and the system dynamics approach of Vennix (1996)), the
facilitator is responsible for using the mapping, simulation or, as in our case, learning
software tool.
The scenario approach aims at overcoming the drawbacks and limitations of
forecasting by providing a structured method to speculate about possible futures. The
value of scenario planning does not stem from its outcomes but rather from the
process of scenario construction itself which stimulates learning. On the other hand,
model construction and manipulation, as well as interactive simulation are also widely
used tools to enhance learning, so that more educated decisions are made.
Furthermore, group model building is a process used to integrate and coordinate
mental models and contexts of individual managers participating in strategy
formulation (Eden and Ackermann, 1998; Vennix, 1996).,
In the proposed methodology, scenarios are built to speculate about the future trends
and competitive forces that shape the external environment and to derive the required
manufacturing performance objectives profiles (required performance with respect to
time). The instantiation/calibration of the generic system dynamics model template of
manufacturing operations, which in effect constitutes a group model-building
exercise, and the simulations with the resulting model are used to understand the
inherent dynamics of the firm’s manufacturing-resource and capability architecture, as
well as to evaluate intended improvement programmes. In this sense, system
dynamics modelling is accomplished in an “interactive” mode to enhance the social
and knowledge processes of strategy formulation (Lane, 1999; Lane, 2000).
In brief, our methodology consists of three learning exercises (LI to L3) structured
into the following stages:
Ll. | LEARNING THE DYNAMICS OF RESOURCE-BASED
MANUFACTURING STRATEGY (CONTEXT INDEPENDENT)
e Exploration of the system dynamics model of manufacturing strategy —
Facilitator-driven.
e Understanding the linkages between system elements - Facilitator-driven.
L2. | LEARNING THE DYNAMICS OF INTERNAL AND EXTERNAL
ENVIRONMENT (CONTEXT SPECIFIC)
A. ASSESSMENT OF EXTERNAL ENVIRONMENT
Definition of planning horizon — Discussion — Agreement.
Identification of future market events at-large — Discussion — Agreement.
Identification of corporate level event s— Discussion — Agreement.
Identification of product level events — Discussion — Agreement.
Construction of required performance objectives profiles (RPOP) — Facilitator
— Discussion — Agreement.
B. ASSESSMENT OF INTERNAL ENVIRONMENT
e For each product, product group, or business unit, assessment of current
manufacturing performance with respect to the manufacturing performance
objectives (cost, flexibility, quality, speed, dependability) — Discussion -
Agreement.
e Identification of manufacturi lated resources — Discussion — Agreement.
e Determination/estimation of contribution of resources to manufacturing
objectives — Discussion — Agreement.
e Identification of manufacturing-related decisions, improvement activities and
processes — Discussion — Agreement.
e Determination/estimation of contribution of decisions, improvement activities
and processes to resource accumulation — Discussion — Agreement.
e Determination of linkages among resources and capabilities — Discussion —
Agreement.
e Calibration of system dynamics model — Facilitator — Discussion — Agreement.
L3. | LEARNING FROM THE RESPONSES TO THE ENVIRONMENTAL
SETTINGS (CONTEXT SPECIFIC)
C. DEVELOPMENT OF MANUFACTURING STRATEGY
e Establishment of improvement projects as sequences of timed activities —
Discussion — Agreement.
e Execution of simulations to construct performance profiles — Facilitator —
Discussion.
e Estimation of effort required for achieving these profiles — Discussion —
Agreement.
e Comparison of performance objectives profiles with the required ones —
Discussion — Agreement.
e Repetition of C for the same scenario or for different scenarios (phase A), if
necessary.
e Repetition of C for alternative assessments of internal environment (phase B),
if necessary.
The three core phases of the approach are discusses in more detail below.
3.2 Assessment of external environment
In this phase scenarios of the external environment are constructed for the planning
horizon, which is typically up to five years. The usual scenario building procedure,
which considers events in the economic, social, technical and demographic spheres is
used (Schoemaker, 1996). (It is obvious that the presentation of the detailed scenario-
construction process is not the purpose of this paper. Thorough insights on this
process can be found in Codet (1987), Schwartz (1996) and Van der Heijden (1996).)
For each scenario, the strategic objectives necessary for achieving competitive
advantage are identified. These are then translated into required timed characteristics
for operations associated to product or product group offerings, and expressed in a 1
to 5 scale (1 = very weak, 2 = weak, 3 = average, 4 = strong, 5 = very strong
requirement) with respect to the performance objectives of cost, flexibility, quality,
dependability and speed.
The SYDMAS learning environment allows required performance objectives profiles
to be defined interactively either by specifying ratings for specific periods, at different
levels of time detail (week, month, 6-month, etc.), or by drawing patterns of evolution
of importance (again within the range of | to 5). Figure 2 shows a screen-shot of the
related interface and a diagram of the input strategic manufacturing profiles for a
specific product group.
BHO we
Fladiity — Depeundatitity
Heawon a] a mPGPee| ROMA] RAUREL AL
ee
ad
Figure 2 The SYDMAS interface for the definition of the required performance
objectives profiles.
3.3 Assessment of internal environment
Once a qualitative assessment of the company’s current manufacturing performance is
accomplished with respect to the performance objectives for the products, product
groups or business units under consideration, the assessment of internal environment
concentrates on the identification and the assessment of manufacturing related
tangible and intangible resources (e.g. packaging machine capacity, workers training
level) which play a significant role in achieving the objectives set in the previous
phase. These resources are listed with an evaluation mark in the range 1 to 5 to
indicate, at the two extremes of the scale, whether they are fully developed strategic
resources (given a rating of 5) or resources which are marginally developed (rating =
1).
Then the res_obj_coeff matrix is constructed. Each element of the matrix indicates the
degree of importance of each resource with respect to each of the five principle
performance objectives. Again, the scale of ratings is between | and 5 (1 = very small
contribution, 2 limited, 3 = average, 4 = important, 5 = decisive resource in
achieving the specific objective).
The next step in this phase involves the identification of the appropriate decision areas
(activities), and the estimation of the influence that each decision and related action
has on each of the previously identified resources (e.g. how the investment on an
advanced scheduling system influences the level of the associated complex scheduling
capability, or how the upgrading of a packaging machine enhances the packaging
capability of the company). Associations between decisions/activities and resources
are tabulated in the act_res_coeff matrix. The scale of ratings is again from | to 5 with
the same meaning as for the contribution of resources to performance objectives. The
same takes place for capabilities (act_cap_coeff). The linkages between resources
and capabilities are also discussed and estimated (res_cap_coeff and cap_res_coeff).
All coefficient matrices are used to calibrate (scale) the system dynamics model in an
interactive fashion by taking into account the current state of the internal environment
and the observed or calculated performance. The resulting model presents a
quantitative estimation of the firm’s manufacturing function’s current operation at the
strategic level. At this stage, the execution of the simulation model provides an
estimation of the projected future performance of the manufacturing function with the
current set and levels of resources, capabilities and activities/policies. In addition, it
provides a picture of the inherent dynamics of the current resource-capability system
with no external disturbances in the form of new policies, improvement programs, etc.
3.4 Devel of f. ing strategy
This phase starts with proposals and discussions on the initiatives that must be
undertaken for achieving the required performance profiles defined for each scenario.
Effectively, it is a process where individual mental models are exposed and
accommodated into a single perspective through discussion and argumentation. The
use of the learning environment facilitates this task as it allows for the immediate
testing of individual assumptions and intuitions. As strategies are formulated,
assumptions on the current state of the manufacturing system may be revised.
Strategies, as timed investment and improvement decisions, may be formulated as a
whole and then tested using SYDMAS, or alternatively, may be decided “on the fly”
as the simulation runs by observing the results of previous decisions. The projected
manufacturing performance is assessed with respect to the performance objectives
profiles defined in phase A. The total effort (investments) to achieve each profile is
calculated automatically based on estimations of activity costs (sum of activity costs).
As in the Activity Based Costing system, for each activity, the cost per unit execution
has to be estimated, after the activities necessary for strategic initiative are
determined. A risk estimate for each strategic initiative is calculated on the basis of
certainty factors defined for the relations/contributions among activities, resources,
capabilities and performance objectives, as well as for the required performance
objectives profiles. Certainty values are collective estimates usually obtained through
discussion and voting. The total certainty (risk) is the product of the certainty factors
of cascaded associations. Simulations can be d for different RPOPs (different
scenarios of the external environment dynamics), or different internal assessments of
all parameter values/assessments, i.e. for decisions, resources and their interrelations,
as well as the relations among resources and performance objectives.
In practice, frequently, the manufacturing strategy process is driven by an “end-
means” rule, that is, it becomes an attempt to close the gap between current and
required performance, which is translated into a quantitative percentage increase (or
decrease) of specific performance objective within a pre-specified time period, e.g.
increase volume flexibility by 10% in the next six months. To get some directions on
how to do this, the manufacturing strategy team, works backwards (returns to
previous phases). First, by considering the appropriate performance metrics
requirements, finds which resources and capabilities contribute to this performance
objective. Then, the amount of change required for each resource to achieve the new
performance objective is estimated. The next question that has to be answered is how
to achieve this change in resources, i.e. which activities should be executed, and at
what rate, for augmenting the resources.
As the relationships among performance objectives, resources and activities are quite
complex inducing dynamic effects, it is important to experiment with different
decision (activity) combinations and timings to see the effects and the results of each
decision setting. As it is shown in the case example provided in the next section,
SYDMAS and its associated process can efficiently support this process.
4. The use of the methodology - An example case
Almost twenty years ago, FOODCO S.A. (the real name of the company is disguised
for confidentiality) was one of the largest cooperative-owned operations in food
processing in Greece producing a relatively constant range of products around two
business units with distinct manufacturing facilities: tomato and canned vegetable
products (tomato paste, peeled whole and diced tomatoes, ketchup, peas, beans,
pickles, peppers, etc.) and potato products (mashed potatoes, frozen fresh fries, etc).
After a period of unsuccessful investments in forward (distribution) and backward
(development and manufacture of basic food processing equipment) of integration, in
the early nineties, the company was in a bad financial and market position. The
intervention of the state resulted in the restructuring of the company and the
rationalisation of its operations. With a leaner structure the company continued to
Operate in increasingly competitive domestic and international markets searching
continuously for a strategy that will guarantee a sustainable growth and prevent it
from eventual financial problems.
Four years ago, the company was a volume producer for the domestic consumer and
catering markets with marginal presence in European and Eastern European markets.
It also acted as a contract manufacturer for private-label products of major domestic
and international supermarket chains. By then, the company operated three business
units, after the addition of a frozen vegetables unit to the tomato and potato ones. At
that time, FOODCO decided to commit resources on its strategy processes and agreed
to adopt (and act as a test-bed for) the methodology and the tool described in this
paper for the formulation of its manufacturing strategy. It should be noted that
managers of the company had been previously exposed to elements of the
methodology and experimented with the basic model in a series of seminars on the
use of system dynamics modelling and simulation in the formulation of strategy.
A team of seven managers was assembled, comprising of all business unit and
functional managers. The whole process was led by an external facilitator, proficient
in the use of the methodology. In a first meeting, the functionalities of SYDMAS
were presented to the team giving particular emphasis on the underlying model.
Then, six daily sessions of scenario building spread over a period of two months
resulted in three concentric sets of scenarios for the external environment. They were
distinguished by the extent of their geographic coverage: Scenarios Balkans,
11
Scenarios Europe and Scenarios Globe. Starting from the global level and considering
the influence of events to smaller of larger spheres of discourse, for each individual
scenario, Required Performance Objective Profiles were constructed. For the shake of
presentation, here we select three scenarios, one from each area, namely Large Balkan
Markets, Clustering Europe and Open USA. The environmental parameters for the
first were summarised in the existence of large low-variety, low-quality markets for
tomato and potato products in the near proximity of the company, whereas for the
second, the formation of a clustering initiative with other domestic food companies to
promote Mediterranean-diet products in the major Western European markets. The
third scenario was triggered by the existence of a sweeping consumer trend towards
Mediterranean-diet products in the USA.
The determination of RPOPs which involved the quantification of performance
objectives proved to be a very time-consuming task as it caused steaming discussions
among the members of the group. The role of the facilitator was to turn their attention
on estimating RPOPs solely on the basis of the scenarios rather than by considering
the current operational and financial state of the company. Discussions and debates
resulted in the definition of RPOPs for the three scenarios:
S1: Large Balkan Markets
Increase capacity in the tomato and potato processing units by 20% in 3 years
Reduce operating cost by 20% in 3 years
Keep incremental pace (5-10%, yearly) in quality and dependability
Keep product range and speed of response to current levels
$2: Clustering Europe
Increase vegetable product range by 100% in 5 years
Increase quality by 50% in 2 years
Increase dependability by 20% in 3 years
Increase speed by 30% in 2 years
Keep annual increase in operating costs below 15%
S3: Open USA
Increase capacity in all units by 20% in 3 years
Reduce operating cost by 10% in 3 years
Increase product range by 40% in 2 years
Increase quality by 30% in 2 years
Increase dependability and speed by 20% in 2 years
Moving into phase B of the methodology, facilitated discussion among the company’s
managers resulted in the identification of three core resources (dedicated high-speed
equipment (R1), flexible equipment (R2), trained operators (R3)) and three basic
capabilities (fast changeover capability (C1), scheduling capability (C2) and special
relationships with suppliers (C3) (in fact, raw material suppliers (small farmers)
worked in the factory for an extra income and are equity holders). Additional
resources and capabilities belonging to the above macro-resources and macro-
capabilities were considered but are not included in the presentation for the shake of
licity and comprehensiveness. The current state of the company’s manufacturing
assets was estimated as:
Total capacity: 1 500 000 units
Product range: 32
Assessment of scheduling software: 1 (ina 1 to 5 scale)
Percentage of trained workers: 20%
Changeover capability: 1 (ina | to 5 scale)
Scheduling capability: 1 (ina | to 5 scale)
The importance factors (contributions) given to the above resources with respect to
the performance objectives were
Cc F D @ Ss
RI 5 1 5 3. 4
R21 5 3 4 2
R31 5 5 5 2
The main activities that influence the above resources’ and capabilities’ accumulation
were identified to be
Al: Increase of capacity
A2: New product introduction
A3: Training of operators
A4: Installation of flexible machinery
AS: — Installation of scheduling software
The contribution of each activity on the accumulation of every resource was estimated
as in the following matrix
Al A2 A3 A4 AS
RI 5 1 1 1 1
R2 1 5 4 5 5
R3 1 4 5 4 3
Similarly, the contribution of each activity on the accumulation of every capability
was estimated as
Al A2 A3 A4 AS
cl 1 3 5 4 4
c2 1 2 5 3 5
cz. 4 1 1 1 1
The resource-capability matrix was formed as
R2 R3
cl -2 5 5
C2. (3 4 3
C3 5 0 -3
The estimations for the performance of the FOODCO’s current manufacturing system
were, in an 1-5 scale, Cost = 3, Flexibility = 2, Dependability = 4, Quality = 3, and
Speed = 3. The assessments for the company’s resource levels were: RI = 4, R2 = 1,
and R3 = 2. For capabilities, Cl = 1, C2 = 1, and C3 = 5. Based on these values, the
model’s scaling coefficients were calculated. The model was calibrated using these
coefficients.
In phase C, in examining how to deal with scenario Large Balkan Markets, the initial
assumption of the team was that a new manufacturing site in a Balkan country would
be necessary for addressing effectively the required performance objectives profiles.
This could very well support the requirement for increased capacity and low cost, but
simulations showed that even with the most optimistic estimates this would take at
least three years to ramp up (mainly to develop a reliable raw materials supply) with a
very high risk (a risk factor of 0.89), especially in terms of achieving the required
performance in quality and dependability. The cost of this effort was estimated in the
region of 2 500 000 Euros. The actual risk of this scenario was even higher since
commitments had to be made at the earliest time possible.
Alternatively, an extension of the existing facilities to accommodate the requirement
of additional capacity seemed more promising with a lower cost. This scenario
exploited the existing capabilities of FOODCO and the already existing inherent
dynamics of the firms manufacturing assets architecture. Different investment and
improvement programs were evaluated. The best program found assumed a
commitment for the extension of capacity, to be made, at the latest, in about 18
months, accompanied by a parallel introduction of a quality improvement program
with emphasis on employee development (to be started at the earliest possible). The
cost of this effort, which satisfied the 20% cost decrease objective, was estimated to
be around | 500 000 Euros, with a risk factor of 0.27.
The second alternative strategy was then chosen as a starting point for assessing an
eventual change in the focus of importance towards the objectives defined for the
scenario Clustering Europe. The internal structure and performance obtained for the
scenario Large Balkan Markets were used as the departure point for assessing further
strategic moves. The simulations showed that the objectives of scenario S2 could not
be achieved with the same operational architecture. Looking into what caused the
strategic inertia, the team started to discuss the linkages between resources and
capabilities, which obviously governed the dynamic behaviour of the system.
Improvement and investment programs that address the requirements of both
scenarios simultaneously were put on the table. In dealing with these scenarios, the
trade-off relationships among resources and capabilities were examined and redefined
to see their effects in SYDMAS. As scenario $2 implied products which required
more flexible and complex handling in all manufacturing activities, to test the
decision of building a new production line for sauces it was assumed that even the
traditional products were treated the same and packaged in plastic vases. This implied
the development of resources and capabilities for storing intermediates. A new
resource, capability and activity were input into SYDMAS and their contribution and
scaling factors were set. In addition, for increasing quality the team examined the
possibility of gradually shifting the incentives to suppliers towards advanced training
for some of them, and additional support in their agricultural activities for the others.
Simulations showed that although this initiative resulted in an initial increase in costs,
the combined objectives of the two scenarios could be well satisfied.
The same process was followed in considering scenario $3, and the team ended up
with a flexible manufacturing strategy as a set of scalable structural relations,
investment activities and operating procedures that can address the requirements of all
three scenarios, if necessary. The main constituent parts of this strategy were: early
development of a new production process that relaxes the current volume-flexibility
trade-off, gradual but consistent redefinition of company’s relations with its suppliers
and establishment of different HR policies. The main SYDMAS interface indicating
the comparative performance assessment of this strategy (for the five performance
objectives) is shown in figure 3. On the left of the screen are sliders and buttons
through which the intensity and frequency of activity execution were set. On the
bottom of the interface are the tables through which the parameters of the model were
input. Buttons were used to navigate to different interfaces, to control the simulations
and to write and display notes on the scenario being executed.
we
1 CASESNEA 2: ASSESSMENID] 9 ASSESHAENTH] 4 AISESOENTH & ASESNENBL
1 Nev production re (es): volumes bity
radsot reuraised
12 Constant investement in traning
3. Constant rote of new product introductions in
VEG unit (both TOM and POT units)
4, Help supplirs in agreutural production
6. Constant training to noreaze quality
TERRES COEF, v
ER RES COEFFIG| [o___~|] Dp /AETRES COeFFIe2]
Figure 3 The main interface of SYDMAS
In reality, the whole manufacturing formulation process took seven months. During
this period, three facilitators were engaged and five scenarios were considered in
detail. The whole process helped the managers of the company to realize what was
possible and what was not possible and to acquire new knowledge in terms of market
potential and different technical and operational systems. Although the company
started to implement with promising success a strategy focusing on the requirements
of the first scenario (Balkan markets), its managers have formed a picture of what will
look like the markets and the firm’s operating environment when similar conditions to
those defined for the other scenarios will arise. In addition, it seems that FOODCO’s
current strategy has the intrinsic flexibility of addressing the requirements of the other
scenarios effectively. More importantly, the use of the learning environment helped
the decision makers of the company to understand, in a holistic way, the dynamics of
the system of which they are part of.
5. Conclusions
In this paper we have presented a novel manufacturing strategy formulation process
that relies on the use of scenarios and system dynamics modelling. We have presented
the logic of the methodology and the structure of the computer based learning
environment that forms its kernel. The use of the methodology and the associated
CBLE has indicated that both can enhance the formulation process addressing and
overcoming drawbacks of other approaches by providing
a means to represent, understand and take into account the structures responsible
for the dynamics of the manufacturing function’s asset system
- a platform for discussion, argumentation and, consequently, knowledge
elicitation and recombination by exposing diverse mental models and
assumptions of stakeholders
- a platform to “visit” diverse future settings of the internal and external
environment of the company.
Although the case presented concerns a company in a relatively mature sector, the
methodology and the tool can be (and have been) easily tailored for more dynamic
sectors where the potential benefits from their use may be higher. Our current efforts
are towards embedding SYDMAS to a collaboration-supporting information system
(Adamides and Karacapilidis, 2005; Karacapilidis and Adamides, 2003) that explicitly
addresses the requirement for IS support of the social dynamics of the strategy
process. Through the implementation of a formal argumentation schema, this will
increase the quality and usefulness of simulations by using the collective knowledge
of all stakeholders of manufacturing strategy in a more complete and efficient manner.
References
Adamides, E.D., 2002, The use of fitness landscape theory and systems dynamics for
the development of manufacturing strategy, In: G. Frizelle and H. Richards, eds.,
Proceedings of the 2002 Conference of the Manufacturing Complexity Network,
Institute for Manufacturing, University of Cambridge, 53-62.
Adamides, ED. and N. Karacapilidis, 2005, Knowledge management and
collaborative model building in the strategy development process, Knowledge and
Process Management, 12(2), 77-88.
Amit, R. and P.J. Schoemaker, 1993, Strategic assets and organizational rent,
Strategic Management Journal 14, 33-46.
Codet, M., 1987, Scenarios and Strategic Management (Butterworth, London).
Dierickx, I. and K. Cool, 1989, Asset stock lation and the inability of
competitive advantage, Management Science 35, 1504-1511.
Eden C. and F. Ackermann, 1998, Making Strategy: The Journey of Strategic
Management (Sage, London).
Ghemawat, P., D.J. Collins, G.P. Pisano and J.W. Rivkin, 2001, Strategy and the
Business Landscape: Core Concepts (Prentice Hall, Upper Saddle River, NJ).
16
Grant, R.M. 1991, The resource-based theory of competitive advantage: Implications
for strategy formulation, California Management Review, Spring 33(2), 114-135.
Groessler, A., 2005, An exploratory system dynamics model of strategic
manufacturing capabilities, Proceedings of the 23" International Conference of the
System Dynamics Society (CD-ROM version).
Hayes, R.H. and S.C. Wheelwright, 1984, Restoring Our Competitive Edge:
Competing through Manufacturing (Wiley, New York).
Hayes, R.H., G.P. Pisano and D.M. Upton, 1996, Strategic Operations: Competing
Through Capabilities (The Free Press, New York).
Hill, T., 2000, Manufacturing Strategy: Text and Cases om edition) (Palgrave, New
York).
Isaacks, W. and P.M. Senge, 1994, Overcoming limits to learning in computer-based
learning environments, In: J.D.W. Morecroft, and J.D. Sterman, eds., Modeling for
Learning Organizations (Productivity Press, Portland, OR) 267-287.
Joseph, A.T., 1999, Formulation of manufacturing strategy, International Journal of
Advanced Manufacturing Technology 15, 522-535.
Karacapilidis, N. and E. Adamides, 2003, Integrating Simulation and Argumentation
in Organizational Decision Making. In V. Palade, R.J. Howlett and L. Jain, eds.,
Knowledge-Based Intelligent Information and Engineering Systems, Proceedings of
the 7th International Conference on Knowledge-Based Intelligent Information and
Engineering Systems (KES 2003), Oxford, Lecture Notes in AI, Vol. 2774, (Springer-
Verlag, Berlin) 107-114.
Lane, D.C., 1999, Social theory and systems dynamics practice, European Journal of
Operational Research, 113, 501-527.
Lane, D.C., 2000, Should system dynamics be described as a “hard” or
“deterministic” systems approach?, Systems Research and Behavioral Science, 17, 3-
Mills, J., K. Platts, and M. Gregory, 1995, A framework for the design of
manufacturing strategy processes, International Journal of Production and Operations
Management 15, 17-49.
Mintzberg, H. and J.A. Waters, 1985, Of strategies deliberate and emergent, Strategic
Management Journal 6, 257-272.
Mintzberg, H. and J. Lampel, 1999, Reflecting on the strategy process, Sloan
Management Review 40 (Spring), 21-30.
Mollona E., 2002, A competence view of firms and resource accumulation systems: A
synthesis of resource-based and evolutionary models of strategy-making, In: J.
Morecroft, , R. Sanchez and A. Heene, eds., Systems Perspectives on Resources,
Capabilities, and Management Processes (Elsevier Science, Oxford) 93-125.
Morecroft, J., 1988, System dynamics and microworlds for policy makers, European
Journal of Operational Research 35, 301-320.
Morecroft, J., 1999, Management attitudes, learning and scale in successful
diversification: A dynamic and behavioural resource system view, Journal of the
Operational Research Society 50, 315-336.
Morecroft, J., 2004, Mental models and learning in system dynamics practice, in:
Pidd, M., ed., Systems Modelling: Theory and Practice, (John Wiley and Sons,
Chichester, UK), 101-126.
Nelson, R.R. and S.G. Winter, 1982, An Evolutionary Theory of Economic Change
(Harvard University Press, Cambridge, MA).
Papert, S., 1980, Mindstorms: Children, Computers, and Powerful Ideas, (Basic
Books, New York).
Platts, K.W. and M.J. Gregory, 1990, Manufacturing audit in the process of strategy
formulation, International Journal of Operations and Production Management 10, 5-
26.
Quezada, L.E., F.M. Cordova, S. Widmer and C. O’Brien, 1999, A methodology for
formulating business strategy in manufacturing firms, International Journal of
Production Economics 60-61, 87-94.
Riis, J. and Smeds, R., 1998, The future of simulation games: issues and challenges,
in: R. Smeds and J.O. Riis, eds., Experimental learning in production management.
The effects of using simulation games in universities and industry (Chapman and
Hall, London) 157-164.
Schoemaker, P. J. H., 1995, Scenario planning: A tool for strategic thinking, Sloan
Management Review 36 (Winter), 25-40.
Schwartz, P. 1996, The Art of the Long View: Planning for the Future in an Uncertain
World (Currency, New York).
Skinner, W., 1969, Manufacturing: Missing link in corporate strategy, Harvard
Business Review 47, 136-145.
Slack, N. and M. Lewis, 2002, Operations Strategy, (Pearson, Harlow).
Smeds, R., 2003, Simulation for accelerated learning and development in industrial
management (Guest editorial), Production Planning and Control 14, 107-110.
Tan, K.H. and K. Platts, 2004a, Connectance model revisited: A tool for
manufacturing objective deployment, Journal of Manufacturing Technology
Management 15, 131-143.
Tan, K.H. and K. Platts, 2004b, Operationalising strategy: Mapping manufacturing
variables, International Journal of Production Economics 89, 379-393.
Taskinen, T., 2003, Improving change management capabilities in manufacturing:
from theory to practice, Production Planning and Control 14, 201-211.
Van der Heijden, K., 1996, Scenarios: The Art of Strategic Conversation (Wiley,
Chichester, UK).
Vennix, J.A.M., 1996, Group Model Building: Facilitating Team Learning Using
System Dynamics (Wiley, Chichester, UK).
Warren, K., 2002, Competitive Strategy Dynamics (Wiley, Chichester, UK).
