AW Gavine
(Royal Armament and Research and Development Establishment)
and
E F Wolstenholme
(University of Bradford Management Centre)
Abstract
Any attempt to impose a computerised information system (CIS) upon
an organisation requires an assessment of its impact in terms of costs,
benefits and procedural change. This paper briefly describes the capacity of
the System Dynamics technique to capture the essence of an organisation's
management structure and to assess, from a system-wide perspective, the
impact of imposing a CIS. The paper employs, as a basis, two case studies set
in a military context and a particular methodology developed with these
applications in mind.
The efficacy of system dynamics in assessing the true impact of CIS on
the enterprise and the user, is appraised, based on a set of independent criteria.
The significance of the methodology for CIS development generally is
considered and encompasses an elaboration of its place in the software life
eycle.
Introduction
Recently, information technology has assumed a rapidly expanding
role in supporting the military commander's decision-making capability and
his exertion of organisational control. Frequently, however, such command,
control and information systems (CCIS) fail to achieve their expectations upon
implementation. Consequent user cynicism is encouraging a more diligent
scrutiny of the objectives, functions and possible design options of a CCIS prior
to its development.
CCIS are increasingly being employed within the military tactical
environment. Procurement of these systems is frequently very complex, and
usually involves several phases each of which can be protracted and expensive.
A crucial task in the early phases of the procurement cycle is to obtain an
estimate of the expected operational benefits and drawbacks. The primary
objective of this study is to investigate the usefulness of the System Dynamics
395
System Dynamics '90
method in assessing the effectiveness of such CCIS. A secondary objective is to
assess the utility of extending the System Dynamics method into other phases of
the procurement life-cycle such as the evaluation of alternative system
designs.
Th th of th
Unlike many traditional hard modelling environments, the aim of a
System Dynamics study extends beyond providing merely a quantitative
description of a system and a simulation of its behaviour within the rigid,
technique-dependent constraints of a goal-directed methodology. System
Dynamics is also a qualitative tool which encourages the participation of all the
relevant actors in a holistic and educative debate and embodies many of the
concepts associated with soft systems thinking. This appraisal of the System
Dynamics technique will be based on a series of criteria having general
applicability, each of which will constitute a measure of the suitability of the
method in evaluating the effectiveness of CCIS.
Application
In order to investigate the generality of the’ System Dynamics
technique, two candidate areas from the Land Systems tactical military
environment were selected for the initial study and, correspondingly, two
parallel study streams were initiated. To ensure that the study had a more
universal relevance, two study streams detailed below were chosen from quite
different but complementary areas.
Battlefield CCIS
A conceptual battlefield CCIS might, if deployed, assist battlegroup
commanders, their staffs, battlegroup elements down to individual vehicles
and associated support elements. The intent is that it will improve the present
command and control process thereby giving commanders and staff more time
for decision making based on accurate, relevant and timely information.
This project is currently in the pre-feasibility phase and a detailed requirement
has yet to be established.
Logistics CCIS
This is a possible logistic CCIS applicable to the higher levels of
battlefield military command. It will support the various logistics cells by
providing common computing facilities such as word processing, message
handling, recording and automatic dissemination of data, data accessing,
order production, calculation aids, ete. This project is already well down the
development path having completed the analysis and logical design phases.
System Dynamics '90
397
The Approach
The means of employing System Dynamics in both application areas
was characterised by 3 stages:-
Stage 1 involves knowledge acquisition through discussion between the
clients and modellers. It attempts to define the scope of the investigation, to
identify real-world symptoms and possible causes and to capture them in the
form of an influence diagram. These diagrams form a structural
representation of the relationships and dependencies between the entities
within the problem space and focus on the behavioural dynamics by means of
an explicit realisation of the feedback processes involved.
Stage 2 involves the superimposition of CCIS representation on the base
model by modelling the effects of CCIS attributes on the environment - not in
their physical manifestation, but in terms of their effects on organisational
activities. In general terms, the CCIS is represented in terms of the quality,
relevance and timeliness of information.
Stage 3 allows the effect of changes in policy or objectives in the
organisation to be explored, with the aim of optimising the application and
employment of CCIS. This may include necessary changes in operating
procedures brought about by adoption of the CCIS and allow comparison of the
benefits and drawbacks of proposed configurations.
This approach is described in greater detail elsewhere (Wolstenholme
1990)
n appli
This study stream involved deriving a sufficiently detailed
understanding of the military logistics organisation in terms of its objectives,
structure, activities and policies, and the determination of organisational
performance measures. A representative activity was selected and the policies
and activities relevant to that area specified in detail. The models were
designed around an acceptable level of aggregation and, in stage 2 of the
methodology, included specification of the principal CCIS attributes such as
comprehensibility, relevance, availability, timeliness and accuracy of
information. The models, written in DYSMAP2, were driven by a selection of
scenarios. The thrust of the study was aimed at providing an holistic
appreciation of the CCIS on the host system in terms of high level performance
measures relating to system achievement. It was concerned with assessing the
need for information as a function of the way in which the information will be
used. A full account of this process is described elsewhere (Watts 1990).
System Dynamics ’90
Th Hi it
This study stream began by attempting to conceptualise and structure
the analysis of the CCIS problem area through discussions with the client. The
model construction and its subsequent development provided a means of
structuring the debate and stimulated insight by study of the model's output. By
employing System Dynamics as a data capture and knowledge structuring aid,
the study stream highlighted some of the soft systems aspects of the method. The
use of influence diagrams as a communications medium was particularly
significant, instigating discussion on modes of functional representation.
The construction of a baseline model, using the STELLA package, was based on
these representations and subsequently enhanced by the superimposition of the
CCIS. Interpretation of the measures of organisational effectiveness as output
during model execution yielded additional insight into the identification of
critical functions and their response under varying patterns of CCIS support. A
full account of this process is described elsewhere (Henderson 1990).
ssessmeni ics Again: Hi
The efficacy and suitability of System Dynamics was appraised by
reference to the following qualitative criteria:-
a. Flexibili iG li
System Dynamics models systems in terms of five distinct conceptual
entities: processes, organisation structure, information structure, strategies
and delays. These entities form a basic primitive set which facilitate easy
translation to System Dynamics modelling constructs. The generic nature of
these elements allows considerable flexibility and generality in model
construction.
In the model of ammunition supply the following elements were
represented. The physical process of interest was the movement of
ammunition. The model displayed the underlying hierarchical structure
associated with military command and incorporated the strategies which
controlled the movement of ammunition. The strategies were effected by
monitoring the physical flows (such as ammunition in transit) and
information flows (such as perceived ammunition stock levels) and executing
the appropriate policies (such as stock or resource allocation). The delays were
employed in simulating the transit of various entities such as supply transport.
In the model of battlegroup functions the primary physical flow
represented the availability and attrition of opposing combat units. The
strategies were effected by monitoring the physical flows (such as number of
available active units within range) and information flows (such as perceived
enemy strength) and executing the appropriate policies (such as advance or
System Dynamics '90
399
"shoot and scoot"), The delays were employed in simulating such variables as
repair times or movements.
The model can be graphically presented in either of two formats... Flow
Diagrams link the building blocks with arrows representing the direction of
flow and are useful in demonstrating the assumptions with reference to the
physical processes. Influence diagrams link the building blocks with arrows
showing the direction of influence of one variable on another and are useful in
determining the underlying feedback structure of the organisation and in
forming a basis for quantitative model development.
b. Ease of Problem Formulation and Development of Models
Within System Dynamics, models are formulated through iterative
interaction with the client. However, the iterations result in the formulation of
both a conceptual model (i.e. in the form of a structured analysis on paper) and
an executable model (normally created using a specialist System Dynamics
tool).
In the case of the logistics model the the System Dynamics concept of
stocks and flows mapped naturally to the physical nature of the application.
The development of the strategies and information sources and sinks was built
on this base model in a relatively straightforward top-down manner.
In the case of the battlefield CCIS the disparate functionality of a
battlegroup was modelled by establishing concepts underlying each separate
function and then defining their interdependency by means of influences.
The ease with which conceptual models are formulated is dependent
upon the investigator's skill at knowledge elicitation, capture and structuring,
upon the enthusiasm and responsiveness of the client and upon the efficacy of
the graphical and other tools employed. In the case of System Dynamics,
influence and flow diagrams provide an accessible and easily intelligible
medium for expression of the colloquy. In similar future studies model
development effort may be mollified by employing a set of guidelines
formulated for this type of problem which will help in overcoming the
inexperience of modellers new to the technology.
The formulation of executable models is eased by selection of an
appropriate dynamic modelling software package which automates many of the
tasks involved in model development and obviates the need for a technical
computer specialist. Accessible, user-friendly packages such as STELLA
which exploit the full range of contemporary MMI capability are particularly
valuable.
400 System Dynamics '90
c. Abili Cope With Complexi
The ability to cope with complex issues is a concern of all
analysis/modelling techniques. The model constructor can quickly lose track
of the interactions of the various elements within the model, and cognitive
limits are soon overrun.
The logistics function is an extremely complex organisation
encompassing many commodities, modes of transport, routes and locations. A
model including detailed representation of the processes associated with all
services and commodities would be expensive to develop, both in man-hours
and hardware requirements. It is likely that the complexity of such a model
would render the interpretation of its behaviour relative to the nature of the
supporting information system extremely difficult. Consequently only a
representative activity of the logistics operation (the ammunition supply
function) was modelled. The impact of the CCIS on this activity was assessed
in relative isolation. The assessment could subsequently be extrapolated to the
whole organisation.
In the case of the battlefield CCIS the resolution of the primitive
functions was only taken as far as necessary to assess the system-wide impact
of a CCIS. The granularity of the model was controlled by appropriate
aggregation of functionality. For example the reinforcement of a force may be
represented as a simple pulse omitting the nature or precise pattern of
reinforcement.
d. Transparency (Clarity) of Assumptions
A client's appreciation of assumptions underlying a model is
dependent upon their perception and sophistication as a model user and the
simplicity of the model construction. The cause and effect linkages employed
in System Dynamics enable the assumptions inherent in the model
formulation to become apparent to the system owner.
In the logistics case the nature of the tool was particularly appropriate
for modelling ammunition movements enabling easy identification of the
storage locations and directions of ammunition movement. Similarly
information flow was easily traced to particular sources and sinks enabling
policies to be identified.
The battlefield CCIS study employed various modelling techniques
which did much to limit the distance of the cognitive horizon, such as
restricting the number of variables, aggregating variables with common
characteristics or limiting the amount of uncertainty incorporated.
System Dynamics '90
ce van. Probl vin;
The System Dynamics approach is aimed at creating change through
understanding. Illumination of the problem area and the facilitation of
insight by means of the process of analysis and model construction is more
important than attempting to provide definitive answers concerning predicted
system behaviour.
For example, during the development of the logistics model the systemic
world view of the System Dynamics model clearly highlighted the
contradictory goals of operational and logistics personnel and the requirement
for policies that balance their needs.
The battlefield CCIS study similarly showed concern for the co-
ordination of the disparate battlegroup functions and their interdependencies
as an expression of the system-wide impact of the CCIS.
f. Realism
System Dynamics attempts to capture the adaptive information
feedback and strategies of organisations. The emphasis is on defining how
organisations contribute to their own problems, rather than blaming external
influences.
Within both CCIS study streams the variables employed had a direct.
and apparent real-world counterpart for example ammunition supply rate in
the logistics model or attrition rate in the battlefield model. No causal links
were included unless they were acknowledged as existing in the real-world
organisation.
The realism of the conclusions drawn is dependent upon the perception
of the relevant actor, whether client or analyst. Realism is approached by
achieving a consensus deriving from the differing viewpoints of all
participants in the system.
g. Knowledge Acquisition
System Dynamics models are formulated through iterative interaction
with the client. A useful starting point is to provide a rudimentary,
unembellished model of the system under investigation which will normally
provoke an immediate and positive response from the client. A suitable model
may act as a germ which will grow and develop as the concepts and structure
underlying the system are exposed. The clarity of these models captures the
knowledge of the client more succinctly than mere verbal expression.
This was illustrated in the logistics study when the demonstration
System Dynamics '90
model provoked a positive and immediate response and stimulated the logistics
system owners to present their requirements in a useful and assimilable form.
h. Interaction and Communication with System Owners
In many systems, the physical processes can be observed by the
modeller, but the control policies and mode of information use must usually be
obtained by direct interaction with the system owners. In the case of live battle
systems, direct observation is unrealistic and the modeller must rely on the
system owners to describe the physical processes as well. It is therefore
particularly vital that in these applications all the actors have a common
medium of expression.
Influence diagrams are easy for a client to understand, and thus
provide a sound medium for communication between client and investigator.
The client's involvement at every step of model formulation and development
increases commitment to the model and encourages a more proactive role.
Clearly, influence diagrams are not the only medium of expression and there
are situations where other devices work better. But there are few such devices
which permit such a closely coupled and immediate relationship between model
and user perception.
i. Ability toc Insigt
One of the most powerful aspects of System Dynamics is that it
encourages systemic interactions to be considered from a holistic perspective.
Participation in the model development process encourages the client to
question pre-conceived ideas and to formulate an original approach.
By studying the CCIS integration with current C2 procedures, the method
can highlight areas where introducing elements of the CCIS might proved
detrimental to the system as a whole. Similarly the method can also be used to
experiment with standing procedures in order to make best use of the CCIS.
In both study streams, the essential systems approach compelled the
system owners to consider the relationships between the system components
and to appreciate the system dependencies.
mn i in: ntitati iteri:
The efficacy and suitability of System Dynamics was appraised by
reference to the following quantitative criteria:-
System Dynamics '90
a. Ease of Learning
Quantitative System Dynamics requires the construction of an
executable model and the modeller must therefore learn a relevant computer
language. The labour involved in translating an influence diagram into a
computer executable model may be assuaged by the use of specialist software
(such as STELLA or DYSMAP). Such packages automate some of the more
routine processes, provide built-in functions and provide debugging aids such
as a variable dependency analyser. However, although such tools may appear
easy to use, care must be taken that the broader objectives of the technique are
understood if misapplication is to be avoided.
The logistics CCIS study stream employed DYSMAP2 run on an IBM PC
which employs as source a segment of code consisting of differential equations
using time-suffixed variables. DYSMAP2 is a compiler-based package which
can slow the turn-around time in performing experiments. The support
documentation is relatively inaccessible and some of the provided functions
rather clumsy.
The battlefield CCIS study stream employed STELLA on a MacIntosh.
By contrast with DYSMAP2, STELLA employs as source an easily constructed
influence diagram from which the equations are automatically derived.
STELLA thus provides a more interactive and immediate interface with the
modeller and enables the modeller to focus on the problem rather than the
software.
Inevitably, such tools impose some constraints on the freedom of action
of the model developer, occasionally compelling some compromise to be made
in model design. They impel the modeller to work within the confines of a
shell and, due to memory constraints, usually provide only limited function
libraries. These problems are not normally encountered when using free-
format high level languages. However the quality of support software is
constantly improving and whilst the employment of a general purpose
language may overcome these constraints, its use greatly reduces modeller
productivity and increases model development time.
b. Data Requirements
In both the study streams the "real-world" nature of the models meant
that most data requirements could be defined in precise and meaningful terms.
A typical System Dynamics model includes relationships and flows for
which data is not available. This may be considered a weakness. However,
System Dynamics is not concerned with accurate prediction and so does not
require high quality data in order to yield executable models; rather it seeks to
capture the broad behavioural dynamics of the relevant functions. The System
System Dynamics ’90
Dynamics paradigm attempts to include all relevant interdependencies, even
if performed imperfectly, rather then exclude some from consideration. By
contrast, the hard systems paradigm encourages the incorporation of only those
relationships for which data is known to be available or can be acquired.
When a study attempts to produce definitive answers, there is often an
unjustified degree of precision implied in the values contained in the output.
There is a natural temptation for the client to place more faith in precise
numerical output. System Dynamics provides output in the form of graphical
traces of the broad behavioural dynamics existent within the model under
investigation. It is far easier for both investigator and client to comprehend
trends and modes of behaviour than it to attempt to interpret a set of numerical
values, with no understanding of how these values were derived or their
fallibility.
c. Ease of Validation
The validation of the CCIS effectiveness in both study streams requires
the following to be established:
Does the stage 1 model exhibit the same behaviour as the real-world
system?
Does the stage 2 model accurately reflect the behaviour of the system
after installation of the CCIS?
Validation in the case of CCIS is therefore difficult. The real world
only exists in time of war and, in any case, no tangible form of the CCIS exists.
Therefore model validity is assessed relative to purpose, the level of client
confidence in the model and the satisfactory explanation of observed systemic
behaviour.
The rational-empiricist epistemology of hard systems thinking applies
a reductionist approach to the modelling of systemic phenomena. System
Dynamics instead strives for an understanding of the structure and broad
behavioural dynamics of complex systems. Attitudes towards data, validation,
accuracy and model use are influenced by these differences and it is
unrealistic to expect such models to give accurate predictions of future states.
Definition of Perf. 1 Effecti M
These study streams have assessed the effectiveness of the CCIS in
terms of its impact on its host environment. System Dynamics attempts to
incorporate subjective measures of intangible concepts. The sensitivity of these
variables to changes in CCIS characteristics can be rapidly assessed by means
of experimentation encouraging the development of effective performance
System Dynamics '90
405
measures,
For example the ammunition profile concept in the logistics CCIS model
arose directly from the analysis of model behaviour. Similarly the "overkill
space” concept in the battlefield CCIS model arose from experimentation with
the model.
e. D i hi
The depth of the analysis can be extended by focussing the study on
particular sub-systems and expanding the corresponding sub-model. The
breadth of analysis can be similarly extended to encompass areas previously
outside the study boundary with co-incident expansion of the model's scope.
The ability of System Dynamics to conveniently aggregate and disaggregate
functions and variables eases this process while maintaining a firmly
orchestrated approach to the study objectives.
In general the depth of analysis achieved depends on the hardness of the
system and the ability of the analyst. In a harder system, the depth of analysis
can be extended to encompass the use of sophisticated control theory and
optimisation methods. In the case of the CCIS models, such a deepening could be
applied to any of the component functions although in some cases with dubious
utility.
f. Ability to Create Insights
System Dynamics models cover a greater scope and manipulate more
variables than is possible within normal human cognitive limits.
Understanding and insight comes from interpretation of results in terms of the
feedback loop structures of models. Observed model behaviour can often
surprise and even be counter-intuitive inciting a closer analysis of the system
structures which is in turn rewarded by additional insight.
g. Ease of Model Development and Analysis
The iterative nature of System Dynamics analysis imposes an
incremental approach to model development. As the understanding of the
system improves, additional areas of importance become apparent.
Consequently, System Dynamics models tend to lead their own development.
The limits to this process are provided, in breadth, by the problem boundaries
and, in depth, by the level of resolution required.
The success of an System Dynamics model leans rather less upon the
availability of empirical data than do other methods. This serves to improve
model flexibility and adaptability and to ease development in intangible areas.
System Dynamics '90
The development of executable models is considerably eased by the
provision of support tools and functions, such as dimension validation routines
and the provision of devices to support sensible structuring of the code
segments. The latter point is particularly relevant in DYSMAP2 where no
structure is imposed on the order of equations.
STELLA has the weakness that no automated dimension checking
‘facilities are provided. This omission can give rise to dimensional imbalance
in development models.
h. Degree of Implementation Achievable
This is high. Ultimately the enhanced understanding achieved
perpetuates the problem solving capability of system owners.
The ease with which System Dynamics models can be constructed,
executed and modified (particularly when using interactive packages) means
that ideas and suggestions from both the analyst and the client can be
implemented rapidly. However, all software packages have limitations which
impel some compromise in representation of real-world features.
For example DYSMAP2 has some limitations with representing
variable transfer times. There is a dearth of variety in the probability
distributions provided. Most important of all the DYSMAP2 restricts user
control over the random number streams provided which can create problems
with comparative assessments of similar runs.
i. Abili Deal with U .
System Dynamics is essentially deterministic normally producing the
same output for every run given the same set of inputs. However, stochastic
elements can be represented within a model, using facilities provided in both
DYSMAP2 and STELLA, enabling variability and uncertainty to be
incorporated.
js ue Ac in indi rr Unit Effor'
The method is very productive in modelling terms, since a small input
of time and effort can lead to meaningful insights. For example, at an early
point in the development of the logistics model, the problems resulting from
resource capacity constraints such as transport were confronted
The method also enables models to be constructed in a short time
(particularly if the modelling software semi-automates some of the model
building procedures) and for little cost. With regard to the particular
application to CCIS assessment, the technical and theoretical base for the
System Dynamics ’90
methodology has been established and, given that a formal set of guidelines can
be produced, model development time will be further reduced.
k. Contribution to Training
Most System Dynamics models can be converted to gaming situations
and used as direct training tools. This is particularly useful when attempting
to analyse the impact of strategy decisions on future outcomes, where the method
can feed back an immediate result. This could be useful in training CCIS
Managers and designers and in demonstrating the consequences of their
actions. However the training aspects were not overtly considered in this
study.
Weaknesses and Strengths
System Dynamics is based on principles of control theory, modelling
cause and effect in terms of feedback loops. This imposes particular structures
on a formulation of the problem area. The resulting viewpoint is adaptable and
incisive but it is not the only view. Many aspects of the CCIS assessment
problem will benefit from an alternative approach to its solution performed in
parallel. Indeed some aspects will require an alternative method of solution:-
those which involve the assessment of specific detailed features; those
embodying detailed technical aspects; those where each entity involves explicit
representation of a large number of attributes within one entity.
The strengths of the method are considerable. It is flexible and
adaptable and able to grapple with complexity. It is a rapid modelling
technique which involves the client fully and which serves as a medium for the
expression of requirements. It is capable of capturing the essence of an
organisation's structure and function for relatively little cost or effort. When
supplemented by other methods for the study of particular detailed areas, the
technique can make a valuable contribution to CCIS assessment.
The CCIS Life-cycle
The life-cycle associated with the procurement of CCIS will
traditionally proceed through several phases - from feasibility through
analysis to design (both logical and physical) and eventually to specification,
development and implementation. The continued justification of any project
depends on a parallel process of benefit/cost assessment at each phase. This
process will also feed back enhancements or amendments to earlier phases for
re-evaluation. This feedback process is illustrated in Fig. 1 below.
System Dynamics ‘90
Specification
Feasibility -—] Analysis F-#} Design [P| Develapment
ra i -_
———__
~t
Fig. 1 The traditional structured evolution of a CCIS
However, no modelling method can be regarded as having a definitive
role to play in this process. Commonly employed methods tend to focus on
specific benefits relating to the CCIS itself and neglect the context of the
underlying application. The weakness of this approach is especially apparent
during the earliest phases of the software life-cycle
The two streams of study indicate that use of this methodology in
parallel with the development process can contribute throughout the earliest
phases of the CCIS life-cycle. In particular, by focussing on the formative
phases of the life-cycle, the ultimate CCIS specification is improved, thereby
improving anticipation of latent or potential problems and lessening the risk of
an expensive retrofit at a later point.
Requirements capture is an ill-structured task which can often result
in a narrowing of perspective because of difficulties in expression. Recently,
the prototyping paradigm has evolved whereby the main analysis and design
principles are applied and a scale model constructed for client assessment.
The System Dynamics modelling approach parallels this process, constructing
a model of the real-world system and incorporating the main features of the
CCIS design for client assessment. The System Dynamics model itself
becomes a component of the requirement.
The study has concentrated on System Dynamics as a means of
assessing the benefits arising from any improvement of operational
effectiveness in the target application area. It has also indicated the potential
of the method in contributing directly to the various phases of the CCIS life cycle
itself. In particular, the battlefield CCIS study stream has shown the benefits of
System Dynamics as a qualitative, participative method, well suited to the ill-
structured problems and ephemeral issues associated with the feasibility or pre-
feasibility phases of system investigation. The logistics CCIS study stream has
demonstrated the utility of capturing the essence of an organisation's
management structure and, by revealing its dynamic behaviour under various
assumptions, assessing the benefits of particular designs.
System Dynamics '90 409
Conclusions
Though all CCIS systems have some degree of similarity, every CCIS
system is different in some way and each possesses unique features.
Consequently CCIS assessment is a complex, multi-faceted and costly process.
The decision support capability required for CCIS assessment must similarly
be capable of complex analysis, have an ability to deal with radically different
types of problem (technical, financial, behavioural, etc.) and be cost-effective
in itself. This implies that ultimately each CCIS assessment exercise requires
a selection of decision support tools to be applied as required. This study has
indicated that System Dynamics provides a valuable addition to the armoury of
weapons available to the decision support analyst.
References
¥, Wolstenholme E. F., Gavine A., Watts K., Henderson S., 1990 "The
Development of a Dynamic Methodology for the Assessment of Computerised
Information Systems.”, International System Dynamics Conference, Boston,
Mass., 1990
2. Watts K., Wolstenholme E. F. 1990 "The Application of a Dynamic
Methodology to Assess the Benefit of a Logistic Information System in
Defence.", International System Dynamics Conference, Boston, Mass., 1990
3. Henderson S., Wolstenholme E. F. 1990 "The Application of a Dynamic
Methodology to Assess the Benefit of a Battlefield Information System.",
International System Dynamics Conference, Boston, Mass., 1990
Copyright © Controller HMSO London 1990.
CC BY-NC-SA 4.0